annotate ITSx @ 5:c711ca9a7f4c draft

Uploaded
author okorol
date Fri, 05 Jun 2015 11:10:49 -0400
parents f82c70f54bd7
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1 #!/usr/bin/perl
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2 # ITSx ITS Extractor
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
3 $app_title = "ITSx -- Identifies ITS sequences and extracts the ITS region";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
4 $app_author = "Johan Bengtsson-Palme et al., University of Gothenburg";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
5 $app_version = "1.0.11";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
6 $app_message = "";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
7 # ----------------------------------------------------------------- #
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
8
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
9 # License information
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
10 $license =
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
11 " ITSx - ITS Extractor -- Identifies ITS sequences and extracts the ITS region\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
12 Copyright (C) 2012-2014 Johan Bengtsson-Palme et al.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
13 \
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
14 This program is free software: you can redistribute it and/or modify\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
15 it under the terms of the GNU General Public License as published by\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
16 the Free Software Foundation, either version 3 of the License, or\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
17 (at your option) any later version.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
18 \
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
19 This program is distributed in the hope that it will be useful,\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
20 but WITHOUT ANY WARRANTY; without even the implied warranty of\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
21 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
22 GNU General Public License for more details.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
23 \
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
24 You should have received a copy of the GNU General Public License\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
25 along with this program, in a file called 'license.txt'\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
26 If not, see: http://www.gnu.org/licenses/.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
27 ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
28
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
29 ## BUGS:
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
30 $bugs = "New features in this version ($app_version):\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
31 - None\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
32 \
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
33 Fixed bugs in this version ($app_version):\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
34 - Fixed a bug causing newline characters to be occasionally skipped in the 'its1.full_and_partial.fasta' FASTA output file when the '--anchor' option was used\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
35 \
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
36 Known bugs in this version ($app_version):\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
37 - None\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
38 ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
39
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
40 ## OPTIONS:
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
41 $options = "\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
42 -i {file} : DNA FASTA input file to investigate\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
43 -o {file} : Base for the names of output file(s)\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
44 -p {directory} : A path to a directory of HMM-profile collections representing ITS conserved regions, default is in the same directory as ITSx itself\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
45 --date {T or F} : Adds a date and time stamp to the output directory, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
46 --reset {T or F} : Re-creates the HMM-database before ITSx is run, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
47
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
48 Sequence selection options:\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
49 -t {character code} : Profile set to use for the search, see the User's Guide (comma-separated), default is all\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
50 -E {value} : Domain E-value cutoff for a sequence to be included in the output, default = 1e-5\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
51 -S {value} : Domain score cutoff for a sequence to be included in the output, default = 0\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
52 -N {value} : The minimal number of domains that must match a sequence before it is included, default = 2\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
53 --selection_priority {sum, domains, eval, score} : Selects what will be of highest priority when determining the origin of the sequence, default is sum\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
54 --search_eval {value} : The E-value cutoff used in the HMMER search, high numbers may slow down the process, cannot be used with the --search_score option, default is 0.01\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
55 --search_score {value} : The score cutoff used in the HMMER search, low numbers may slow down the process, cannot be used with the --search_eval option, default is to used E-value cutoff, not score\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
56 --allow_single_domain {e-value,score or F} : Allow inclusion of sequences that only find a single domain, given that they meet the given E-value and score thresholds, on with parameters 1e-9,0 by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
57 --allow_reorder {T or F} : Allows profiles to be in the wrong order on extracted sequences, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
58 --complement {T or F} : Checks both DNA strands against the database, creating reverse complements, on (T) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
59 --cpu {value} : the number of CPU threads to use, default is 1\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
60 --multi_thread {T or F} : Multi-thread the HMMER-search, on (T) if number of CPUs (--cpu option > 1), else off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
61 --heuristics {T or F} : Selects whether to use HMMER's heuristic filtering, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
62
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
63 Output options:\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
64 --summary {T or F} : Summary of results output, on (T) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
65 --graphical {T or F} : 'Graphical' output, on (T) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
66 --fasta {T or F} : FASTA-format output of extracted ITS sequences, on (T) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
67 --preserve {T or F} : Preserve sequence headers in input file instead of printing out ITSx headers, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
68 --save_regions {SSU,ITS1,5.8S,ITS2,LSU,all,none} : A comma separated list of regions to output separate FASTA files for, 'ITS1,ITS2' by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
69 --anchor {integer or HMM} : Saves an additional number of bases before and after each extracted region. If set to 'HMM' all bases matching the corresponding HMM will be output, default = 0\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
70 --only_full {T or F} : If true, output is limited to full-length regions, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
71 --partial {integer} : Saves additional FASTA-files for full and partial ITS sequences longer than the specified cutoff, default = 0 (off)\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
72 --concat {T or F} : Saves a FASTA-file with concatenated ITS sequences (with 5.8S removed), off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
73 --minlen {integer} : Minimum length the ITS regions must be to be outputted in the concatenated file (see above), default = 0\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
74 --positions {T or F} : Table format output containing the positions ITS sequences were found in, on (T) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
75 --table {T or F} : Table format output of sequences containing probable ITS sequences, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
76 --not_found {T or F} : Saves a list of non-found entries, on (T) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
77 --detailed_results {T or F} : Saves a tab-separated list of all results, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
78 --truncate {T or F} : Truncates the FASTA output to only contain the actual ITS sequences found, on (T) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
79 --silent {T or F} : Supresses printing progress info to stderr, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
80 --graph_scale {value} : Sets the scale of the graph output, if value is zero, a percentage view is shown, default = 0\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
81 --save_raw {T or F} : Saves all raw data for searches etc. instead of removing it on finish, off (F) by default\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
82
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
83 -h : displays this help message\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
84 --help : displays this help message\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
85 --bugs : displays the bug fixes and known bugs in this version of ITSx\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
86 --license : displays licensing information\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
87 ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
88
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
89
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
90 ## Print title message
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
91 print STDERR "$app_title\nby $app_author\nVersion: $app_version\n$app_message";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
92 print STDERR "-----------------------------------------------------------------\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
93
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
94 ## Setup default variable values
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
95 use List::Util qw(first max maxstr min minstr reduce shuffle sum);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
96
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
97 $bindir = $0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
98 $bindir =~ s/_x//;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
99 $input = "";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
100 $output = "ITSx_out";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
101 $hmmscan = "";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
102 $profileDB = "$bindir\_db/HMMs";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
103 $type = "all";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
104 $E = 1e-5;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
105 $S = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
106 $N = 2;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
107 $priority = "sum";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
108 $search_eval = 0.01;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
109 $search_score = "";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
110 $allow_single_E = 1e-9;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
111 $allow_single_score = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
112 #$allow_single_E = -1; # Turns off single-domain matching by E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
113 #$allow_single_score = 0; # Turns off single-domain matching by score
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
114 $allow_reorder = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
115 $complement = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
116 $cpu = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
117 $multi_thread = "unset";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
118 $heuristics = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
119 $out_sum = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
120 $out_graph = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
121 $out_fasta = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
122 $out_preserve = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
123 $out_ssu = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
124 $out_its1 = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
125 $out_its2 = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
126 $out_58S = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
127 $out_lsu = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
128 $out_pos = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
129 $out_table = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
130 $out_not = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
131 $out_date = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
132 $out_joined = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
133 $out_results = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
134 $out_partial = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
135 $out_concat = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
136 $concat_minlen = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
137 $truncate = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
138 $anchor = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
139 $only_full = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
140 $graph_scale = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
141 $debug = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
142 $reset = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
143
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
144 ## Read command-line options
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
145 for ($i = 0; $i <= scalar(@ARGV); $i++) { # Goes through the list of arguments
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
146 $arg = @ARGV[$i]; # Stores the current argument in $arg
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
147
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
148 if ($arg eq "-i") { # Read input files from -i flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
149 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
150 $input = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
151 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
152 if ($arg eq "-o") { # Read output files from -o flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
153 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
154 $output = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
155 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
156 if ($arg eq "-p") { # Read profile database from -p flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
157 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
158 $profileDB = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
159 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
160 if ($arg eq "--hmmscan") { # Read pre-computed hmmscan output file from --hmmscan flag ('undocumented' feature)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
161 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
162 $hmmscan = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
163 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
164 if ($arg eq "--date") { # Determine whether or not to add a date stamp based on the --date flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
165 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
166 if (substr(@ARGV[$i],0,1) =~ m/^[Ff0]/) { # Check if argument begins with "F", "f", or "0"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
167 $out_date = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
168 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
169 $out_date = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
170 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
171 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
172 if ($arg eq "--reset") { # Reset HMM database?
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
173 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
174 if (substr(@ARGV[$i],0,1) =~ m/^[Ff0]/) { # Check if argument begins with "F", "f", or "0"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
175 $reset = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
176 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
177 $reset = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
178 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
179 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
180
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
181 if ($arg eq "-t") { # Select what types of ITSs to look for using the -t flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
182 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
183 $type = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
184 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
185 if ($arg eq "-E") { # Set the E-value cutoff using the -E flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
186 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
187 $E = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
188 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
189 if ($arg eq "-S") { # Set the score cutoff using the -S flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
190 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
191 $S = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
192 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
193 if ($arg eq "-N") { # Set the number of found domains cutoff using the -N flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
194 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
195 $N = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
196 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
197 if ($arg eq "--selection_priority") { # Set how to order the ITS types using the --selection_priority flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
198 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
199 $priority = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
200 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
201 if ($arg eq "--search_eval") { # Set the E-value cutoff for the HMMER search using the --search_eval flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
202 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
203 $search_eval = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
204 $search_score = ""; # Turns off score cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
205 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
206 if ($arg eq "--search_score") { # Set the score cutoff for the HMMER search using the --search_score flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
207 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
208 $search_score = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
209 $search_eval = ""; # Turns off E-value cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
210 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
211 if ($arg eq "--allow_single_domain") { # Determine whether or not to allow single domain matches based on the --allow_single_domain flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
212 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
213 if (substr(@ARGV[$i],0,1) =~ m/^[Ff0]/) { # Check if argument begins with "F", "f", or "0"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
214 $allow_single_E = -1; # Turns off single-domain matching by E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
215 $allow_single_score = 0; # Turns off single-domain matching by score
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
216 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
217 ($allow_single_E,$allow_single_score) = split(',',@ARGV[$i]); # Turns on single-domain matching, assigning the first given value as the E-value cutoff, and the second as score cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
218 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
219 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
220 if ($arg eq "--allow_reorder") { # Determine whether or not to allow the domains to be in the wrong order based on the --allow_reorder flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
221 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
222 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
223 $allow_reorder = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
224 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
225 $allow_reorder = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
226 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
227 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
228 if ($arg eq "--complement") { # Determine whether or not to scan the complementary strand of the input file based on the --complement flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
229 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
230 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
231 $complement = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
232 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
233 $complement = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
234 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
235 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
236 if ($arg eq "--cpu") { # Set the number of CPUs to use based on the --cpu flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
237 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
238 $cpu = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
239 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
240 if ($arg eq "--multi_thread") { # Determine whether or not to multi-thread the HMMER step based on the --multi_thread flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
241 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
242 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
243 $multi_thread = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
244 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
245 $multi_thread = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
246 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
247 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
248 if ($arg eq "--heuristics") { # Determine whether or not to use HMMER's heuristic filtering based on the --heuristics flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
249 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
250 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
251 $heuristics = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
252 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
253 $heuristics = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
254 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
255 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
256
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
257 if ($arg eq "--summary") { # Determine whether or not to output a summary based on the --summary flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
258 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
259 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
260 $out_sum = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
261 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
262 $out_sum = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
263 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
264 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
265 if ($arg eq "--graphical") { # Determine whether or not to output a graphical representation of matches based on the --graphical flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
266 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
267 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
268 $out_graph = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
269 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
270 $out_graph = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
271 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
272 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
273 if ($arg eq "--detailed_results") { # Determine whether or not to output a detailed results list, based on the --detailed_results flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
274 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
275 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
276 $out_results = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
277 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
278 $out_results = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
279 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
280 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
281 if ($arg eq "--partial") { # Set the full-and-partial cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
282 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
283 $out_partial = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
284 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
285 if ($arg eq "--anchor") { # Set the length of the sequence "anchors"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
286 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
287 $anchor = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
288 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
289 if ($arg eq "--only_full") { # Output only full-length regions
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
290 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
291 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
292 $only_full = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
293 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
294 $only_full = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
295 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
296 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
297 if ($arg eq "--save_regions") { # Determine which regions to output FASTA files for based on the --save_regions flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
298 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
299 @save_regions = split(',',uc(@ARGV[$i]));
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
300 $out_ssu = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
301 $out_its1 = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
302 $out_its2 = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
303 $out_58S = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
304 $out_lsu = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
305 foreach $save_region (@save_regions) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
306 if ($save_region eq "SSU") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
307 $out_ssu = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
308 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
309 if ($save_region eq "ITS1") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
310 $out_its1 = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
311 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
312 if ($save_region eq "5.8S") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
313 $out_58S = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
314 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
315 if ($save_region eq "ITS2") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
316 $out_its2 = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
317 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
318 if ($save_region eq "LSU") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
319 $out_lsu = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
320 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
321 if ($save_region eq "ALL") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
322 $out_ssu = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
323 $out_its1 = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
324 $out_its2 = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
325 $out_58S = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
326 $out_lsu = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
327 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
328 if ($save_region eq "NONE") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
329 $out_ssu = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
330 $out_its1 = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
331 $out_its2 = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
332 $out_58S = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
333 $out_lsu = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
334 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
335 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
336 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
337 if ($arg eq "--positions") { # Determine whether or not to output a positions file based on the --positions flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
338 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
339 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
340 $out_pos = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
341 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
342 $out_pos = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
343 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
344 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
345 if ($arg eq "--concat") { # Determine whether or not to output a concatednated ITS1 + ITS2 file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
346 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
347 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
348 $out_concat = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
349 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
350 $out_concat = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
351 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
352 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
353 if ($arg eq "--minlen") { # Set the min length of the combined ITS1 and ITS2 sequences for concatenation
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
354 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
355 $concat_minlen = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
356 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
357 if ($arg eq "--fasta") { # Determine whether or not to output FASTA-files based on the --fasta flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
358 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
359 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
360 $out_fasta = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
361 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
362 $out_fasta = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
363 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
364 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
365 if ($arg eq "--preserve") { # Determine whether or not to preserve FASTA-headers based on the --preserve flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
366 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
367 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
368 $out_preserve = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
369 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
370 $out_preserve = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
371 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
372 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
373 if ($arg eq "--joined") { # Determine whether or not to output a FASTA-file containing ALL sorts of output sequences (for debugging)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
374 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
375 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
376 $out_joined = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
377 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
378 $out_joined = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
379 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
380 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
381 if ($arg eq "--table") { # Determine whether or not to output tables of all potential matches based on the --table flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
382 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
383 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
384 $out_table = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
385 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
386 $out_table = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
387 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
388 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
389 if ($arg eq "--not_found") { # Determine whether or not to output a list of sequences that are not ITSs based on the --not_found flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
390 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
391 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
392 $out_not = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
393 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
394 $out_not = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
395 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
396 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
397 if ($arg eq "--silent") { # Determine whether or not to output anything to the screen based on the --silent flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
398 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
399 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
400 $silent = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
401 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
402 $silent = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
403 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
404 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
405 if ($arg eq "--graph_scale") { # Set the scale of the graphical output based on the --graph_scale flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
406 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
407 $graph_scale = @ARGV[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
408 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
409 if ($arg eq "--save_raw") { # Determine whether or not to save all the raw intermediate data based on the --save_raw flag
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
410 $i++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
411 if (substr(@ARGV[$i],0,1) =~ m/^[Tt1]/) { # Check if argument begins with "T", "t", or "1"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
412 $save_raw = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
413 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
414 $save_raw = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
415 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
416 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
417
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
418 ## If "-h" or "--help" are among the options, output usage data and options
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
419 if (($arg eq "-h") || ($arg eq "--help")) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
420 print "Usage: ITSx -i <input file> -o <output file>\nOptions:$options";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
421 print "-----------------------------------------------------------------\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
422 exit; # Exit ITSx
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
423 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
424
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
425 ## If "--bugs" is among the options, output bugs and features information
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
426 if ($arg eq "--bugs") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
427 print "$bugs\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
428 exit; # Exit ITSx
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
429 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
430
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
431 ## If "--license" is among the options, output license information
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
432 if ($arg eq "--license") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
433 print "$license\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
434 exit; # Exit ITSx
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
435 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
436
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
437 if ($arg eq "--debug") { # Run ITSx in debug mode
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
438 $debug = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
439 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
440 if ($arg eq "--pipeline") { # Run ITSx in pipeline mode
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
441 $pipeline = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
442 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
443 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
444
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
445 ## Setup some variables dependent on input
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
446
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
447 if ($multi_thread eq "unset") { # If the multi-thread option is not set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
448 if ($cpu > 1) { # Then if the number of CPUs used > 1, then multi-thread HMMER searches
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
449 $multi_thread = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
450 } else { # Else, run HMMER searches sequentially on one CPU
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
451 $multi_thread = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
452 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
453 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
454
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
455 if ($hmmscan ne "") { # If a pre-computed hmmscan output is supplied
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
456 $output = $hmmscan; # Then set the base of the output directory name to be the same as that hmmscan output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
457 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
458
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
459 ## Check for binaries
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
460
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
461 chomp($path = `which hmmpress`); # Get the path for hmmpress
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
462 if ($path eq "") { # If the path is empty, then show an error message and exit ITSx
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
463 print STDERR "FATAL ERROR :: Could not locate HMMER binaries! It seems that hmmpress is not installed properly.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
464 Consult the manual for installation instructions. Note that HMMER3 is required. Previous HMMER-versions will not work.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
465 This error is fatal, and ITSx will now abort.\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
466 print STDERR "-----------------------------------------------------------------\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
467 exit;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
468 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
469
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
470 chomp($path = `which hmmscan`); # Get the path for hmmscan
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
471 if ($path eq "") { # If the path is empty, then show an error message and exit ITSx
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
472 print STDERR "FATAL ERROR :: Could not locate HMMER binaries! It seems that hmmscan is not installed properly.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
473 Consult the manual for installation instructions. Note that HMMER3 is required. Previous HMMER-versions will not work.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
474 This error is fatal, and ITSx will now abort.\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
475 print STDERR "-----------------------------------------------------------------\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
476 exit;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
477 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
478
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
479
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
480 ## Check for database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
481 chomp($errormsg = `ls $profileDB* 2>&1 1>/dev/null`); # Get the error msg when looking for the profile database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
482 if (substr($errormsg,0,4) eq "ls: ") { # If the error message begins with "ls: ", then show an error message and exit ITSx
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
483 print STDERR "FATAL ERROR :: The specified profile database could not be found.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
484 Consult the manual for installation instructions.\
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
485 This error is fatal, and ITSx will now abort.\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
486 print STDERR "-----------------------------------------------------------------\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
487 exit;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
488 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
489
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
490 if ($pipeline == 0) { # If ITSx is not run in pipeline mode (i.e. from ITSx)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
491 if ($out_date == 1) { # If a date and time stamp should be supplied
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
492 ($sec,$min,$hour,$mday,$mon,$year,$wday,$yday,$isdst) = localtime(time); # Get the date and time
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
493 $year = $year + 1900; # Format the year
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
494 $mon = $mon + 1; # Format the month
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
495 if ($mon < 10) { # Add a zero to the month, if needed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
496 $mon = "0" . $mon;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
497 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
498 if ($mday < 10) { # Add a zero to the day, if needed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
499 $mday = "0" . $mday;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
500 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
501 if ($hour < 10) { # Add a zero to the hour, if needed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
502 $hour = "0" . $hour;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
503 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
504 if ($min < 10) { # Add a zero to the minute, if needed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
505 $min = "0" . $min;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
506 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
507 $outputDate = ".$year\-$mon\-$mday\_$hour.$min"; # Create a date and time stamp
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
508 $outputDate =~ s./.-.g; # Remove any potential slashes in the output name (as this will confuse ITSx's file naming)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
509 $output = $output . $outputDate; # Add the date and time stamp top the output base name
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
510 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
511 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
512
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
513 $tempDir = "ITSx_temp_directory__$output"; # Setup a temporary directory variable
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
514 $tempDir =~ s./.-.g; # Remove any potential slashes in the output name (as this will confuse ITSx's file naming)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
515
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
516 if ($pipeline == 0) { # If not running in pipeline mode
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
517 ## Create a summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
518 if ($out_sum == 1) { # If summary output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
519 $now = localtime; # Get the current time
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
520 open (SUMMARY, ">$output.summary.txt"); # Create the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
521 print SUMMARY "ITSx run started at $now.\n"; # Output the starting time for the analysis
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
522 print SUMMARY "-----------------------------------------------------------------\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
523 close (SUMMARY); # Close summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
524 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
525 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
526
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
527 ## Create a temporary directory for ITSx
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
528 if ($pipeline == 0) { # If ITSx is not run in pipeline mode (i.e. from ITSx)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
529 `mkdir $tempDir 2> /dev/null`; # Create a temporary directory
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
530 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
531
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
532
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
533 ## Prepare profile database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
534 ## Get the current time and output info message
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
535 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
536 if ($silent == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
537 print STDERR "$now : Preparing HMM database (should be quick)...\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
538 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
539
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
540 ## Setup profile index
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
541 %profileIndex = {};
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
542
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
543 # A = Alveolata
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
544 # B = Bryophyta
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
545 # C = Bacillariophyta
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
546 # D = Amoebozoa
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
547 # E = Euglenozoa
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
548 # F = Fungi
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
549 # G = Chlorophyta (green algae)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
550 # H = Rhodophyta (red algae)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
551 # I = Phaeophyceae (brown algae)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
552 # L = Marchantiophyta (liverworts)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
553 # M = Metazoa
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
554 # N = Microsporidia
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
555 # O = Oomycota
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
556 # P = Haptophyceae (prymnesiophytes)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
557 # Q = Raphidophyceae
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
558 # R = Rhizaria
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
559 # S = Synurophyceae
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
560 # T = Tracheophyta (higher plants)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
561 # U = Eustigmatophyceae
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
562 # X = Apusozoa
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
563 # Y = Parabasalia
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
564
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
565 $profileIndex{"A"} = "alveolates";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
566 $profileIndex{"B"} = "bryophyta";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
567 $profileIndex{"C"} = "bacillariophyta";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
568 $profileIndex{"D"} = "amoebozoa";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
569 $profileIndex{"E"} = "euglenozoa";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
570 $profileIndex{"F"} = "fungi";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
571 $profileIndex{"G"} = "chlorophyta";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
572 $profileIndex{"H"} = "rhodophyta";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
573 $profileIndex{"I"} = "phaeophyceae";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
574 $profileIndex{"J"} = "undefined";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
575 $profileIndex{"K"} = "undefined";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
576 $profileIndex{"L"} = "marchantiophyta";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
577 $profileIndex{"M"} = "metazoa";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
578 $profileIndex{"N"} = "microsporidia";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
579 $profileIndex{"O"} = "oomycota";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
580 $profileIndex{"P"} = "haptophyceae";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
581 $profileIndex{"Q"} = "raphidophyceae";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
582 $profileIndex{"R"} = "rhizaria";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
583 $profileIndex{"S"} = "synurophyceae";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
584 $profileIndex{"T"} = "tracheophyta";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
585 $profileIndex{"U"} = "eustigmatophyceae";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
586 $profileIndex{"V"} = "undefined";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
587 $profileIndex{"W"} = "undefined";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
588 $profileIndex{"X"} = "apusozoa";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
589 $profileIndex{"Y"} = "parabasalia";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
590 $profileIndex{"Z"} = "undefined";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
591
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
592
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
593 @profileList = split(',',uc($type)); # Get the list of profile types
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
594 foreach $entry (@profileList) { # Go through the entered types
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
595 if (($entry eq "ALL") || ($entry eq ".")) { # If "all" among the entries
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
596 push(@profileSet,"A"); # Add the alveolates profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
597 push(@profileSet,"B"); # Add the bryophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
598 push(@profileSet,"C"); # Add the bacillariophyta profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
599 push(@profileSet,"D"); # Add the amoebozoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
600 push(@profileSet,"E"); # Add the euglenozoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
601 push(@profileSet,"F"); # Add the fungi profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
602 push(@profileSet,"G"); # Add the green algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
603 push(@profileSet,"H"); # Add the red algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
604 push(@profileSet,"I"); # Add the brown algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
605 push(@profileSet,"L"); # Add the liverworts profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
606 push(@profileSet,"M"); # Add the metazoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
607 #push(@profileSet,"N"); # Add the microsporidia profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
608 push(@profileSet,"O"); # Add the oomycetes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
609 push(@profileSet,"P"); # Add the prymnesiophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
610 push(@profileSet,"Q"); # Add the raphidophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
611 push(@profileSet,"R"); # Add the rhizaria profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
612 push(@profileSet,"S"); # Add the synurophyceae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
613 push(@profileSet,"T"); # Add the tracheophyta (higher plants) profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
614 push(@profileSet,"U"); # Add the eustigmatophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
615 #push(@profileSet,"X"); # Add the apusozoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
616 #push(@profileSet,"Y"); # Add the parabasalia profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
617 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
618 if (length($entry) == 1) { # If the name has only one character
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
619 if ($entry =~ m/[ABCDERFHILMNOPQRSTUXY]/) { # If the selected set exists
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
620 push(@profileSet,$entry); # Add the selected profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
621 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
622 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
623 if ($entry =~ m/ALVEOL/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
624 push(@profileSet,"A"); # Add the alveolates profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
625 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
626 if ($entry =~ m/BRYO/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
627 push(@profileSet,"B"); # Add the bryophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
628 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
629 if ($entry =~ m/MOSS/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
630 push(@profileSet,"B"); # Add the bryophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
631 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
632 if ($entry =~ m/BACILL/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
633 push(@profileSet,"C"); # Add the bacillariophyta profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
634 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
635 if ($entry =~ m/DIATOM/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
636 push(@profileSet,"C"); # Add the bacillariophyta profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
637 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
638 if ($entry =~ m/AMOEB/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
639 push(@profileSet,"D"); # Add the amoebozoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
640 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
641 if ($entry =~ m/EUGLE/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
642 push(@profileSet,"E"); # Add the euglenozoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
643 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
644 if ($entry =~ m/FUNG/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
645 push(@profileSet,"F"); # Add the fungi profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
646 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
647 if ($entry =~ m/GREEN/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
648 push(@profileSet,"G"); # Add the green algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
649 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
650 if ($entry =~ m/CHLORO/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
651 push(@profileSet,"G"); # Add the green algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
652 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
653 if ($entry =~ m/RED-AL/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
654 push(@profileSet,"H"); # Add the red algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
655 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
656 if ($entry =~ m/RHODO/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
657 push(@profileSet,"H"); # Add the red algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
658 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
659 if ($entry =~ m/BROWN/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
660 push(@profileSet,"I"); # Add the brown algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
661 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
662 if ($entry =~ m/PHAEOP/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
663 push(@profileSet,"I"); # Add the brown algae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
664 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
665 if ($entry =~ m/LIVER/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
666 push(@profileSet,"L"); # Add the liverworts profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
667 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
668 if ($entry =~ m/MARCH/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
669 push(@profileSet,"L"); # Add the liverworts profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
670 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
671 if ($entry =~ m/METAZ/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
672 push(@profileSet,"M"); # Add the metazoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
673 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
674 if ($entry =~ m/ANIMAL/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
675 push(@profileSet,"M"); # Add the metazoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
676 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
677 if ($entry =~ m/MICROSPOR/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
678 push(@profileSet,"N"); # Add the microsporidia profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
679 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
680 if ($entry =~ m/OOMYC/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
681 push(@profileSet,"O"); # Add the oomycetes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
682 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
683 if ($entry =~ m/PRYMN/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
684 push(@profileSet,"P"); # Add the prymnesiophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
685 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
686 if ($entry =~ m/HAPTO/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
687 push(@profileSet,"P"); # Add the prymnesiophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
688 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
689 if ($entry =~ m/RAPHID/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
690 push(@profileSet,"Q"); # Add the raphidophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
691 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
692 if ($entry =~ m/RHIZA/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
693 push(@profileSet,"R"); # Add the rhizaria profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
694 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
695 if ($entry =~ m/SYNUR/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
696 push(@profileSet,"S"); # Add the synurophyceae profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
697 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
698 if ($entry =~ m/TRACHE/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
699 push(@profileSet,"T"); # Add the tracheophyta profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
700 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
701 if ($entry =~ m/PLANTS/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
702 push(@profileSet,"T"); # Add the tracheophyta profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
703 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
704 if ($entry =~ m/EUSTIG/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
705 push(@profileSet,"U"); # Add the eustigmatophytes profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
706 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
707 if ($entry =~ m/APUSO/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
708 push(@profileSet,"X"); # Add the apusozoa profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
709 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
710 if ($entry =~ m/PARAB/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
711 push(@profileSet,"Y"); # Add the parabasalia profiles to the investigation set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
712 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
713 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
714 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
715 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
716
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
717 foreach $set (@profileSet) { # For each set of profiles in the the full profile set for investigation
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
718 $hmmPath = $profileDB . "/" . $set . ".hmm"; # Determine the path to the HMM-file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
719 chomp($modelCount = `grep -c "//" $hmmPath`); # Count the number of models in the HMM-file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
720 push(@modelCount,$modelCount); # Add the number of models in this HMM-file to the list of model counts
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
721 if ($reset == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
722 `rm -f $hmmPath.h3* 2> /dev/null`; # Delete old HMM-files
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
723 `hmmpress $hmmPath 2> /dev/null`; # Prepare the HMM-file for searching
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
724 ## Redirecting stderr is a quick and dirty solution to get rid of the messages... Could be made more elegant
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
725 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
726 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
727
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
728 ## Clean-up input files and create complementary strand if needed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
729 ## Get the current time and output an info message
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
730 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
731 if ($silent == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
732 print STDERR "$now : Checking and handling input sequence data (should not take long)...\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
733 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
734
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
735 ## Open the summary file for writing
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
736 if ($out_sum == 1) { # If summary output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
737 open (SUMMARY, ">>$output.summary.txt"); # Append to the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
738 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
739
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
740 if ($input ne "") { # If an input file is given
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
741 ## Read from file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
742 open (SEQUENCES, $input); # Open the input file for reading
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
743 open (MAIN, ">$tempDir/main.fasta"); # Create a temporary file for storing the cleaned sequences representing the main strand
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
744 open (COMPLEMENT, ">$tempDir/complement.fasta"); # Create a temporary file for storing the cleaned sequences representing the complementary strand
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
745 $inputSequenceCount = 0; # Reset input sequence counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
746 while ($sequence = <SEQUENCES>) { # Repeat for every line in the input file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
747 chomp($sequence); # Truncate any potential line feeds
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
748 $sequence =~ tr/\r\n//d; # Remove all carriage return and new line characters
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
749 if (substr($sequence,0,1) eq ">") { # If a new FASTA entry is found in the input file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
750 $inputSequenceCount++; # Add one to the input sequence counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
751 print MAIN $mainSeq . "\n"; # Write the previous main DNA sequence to the main sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
752 print MAIN $sequence . " main\n"; # Write the definition line of the new sequence to the main sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
753 $sequenceDB{"$sequenceID"} = $mainSeq; # Add sequence to sequence database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
754 $headers{"$sequenceID"} = $header; # Add the header to the header database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
755 push(@sequenceOrder,$sequenceID); # Add this sequence ID to the ordered list of sequences
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
756 ($sequenceID) = split(" ",substr($sequence,1));
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
757 $header = $sequence; # Save the sequence header
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
758 $mainSeq = ""; # Empty the main sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
759 if ($complement == 1) { # If the complementary file should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
760 $complementSeq = reverse($complementSeq); # Reverse the complementary DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
761 print COMPLEMENT $complementSeq . "\n"; # Write the previous complementary DNA sequence to the complementary sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
762 print COMPLEMENT $sequence . " complement\n"; # Write the definition line of the new sequence to the complement sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
763 $complementSeq = ""; # Empty the complementary sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
764 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
765 } else { # If this line is just a continuation of the current DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
766 $mseq = $sequence; # Store this part of the DNA sequence in the intermediate varaible $mseq
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
767 $mseq =~ s/[ .-]//g; # Remove any gap characters present in this sequence part (good if input was an alignment)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
768 $mseq =~ tr/[a-z]/[A-Z]/; # Make all letters uppercase
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
769 $mseq =~ s/[^A-Z]//g; # Remove all non-alphabetic characters
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
770 $mseq =~ tr/U/T/; # Exchanges U:s for T:s (Uracil to Thymine, good if input was RNA sequence)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
771 $mainSeq = $mainSeq . $mseq; # Add the intermediate DNA sequence to the end of the main DNA sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
772
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
773 if ($complement == 1) { # If the complementary file should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
774 $cseq = $sequence; # Store this part of the DNA sequence in the intermediate varaible $cseq
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
775 $cseq =~ s/[ .-]//g; # Remove any gap characters present in this sequence part (good if input was an alignment)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
776 $cseq =~ tr/[a-z]/[A-Z]/; # Make all letters uppercase
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
777 $cseq =~ s/[^A-Z]//g; # Remove all non-alphabetic characters
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
778 $cseq =~ tr/ACGTURYSWKMBDHVN/TGCAAYRWSMKVHDBN/; # Replace all characters with its complementary base
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
779 $complementSeq = $complementSeq . $cseq; # Add the intermediate DNA sequence to the end of the complementary DNA sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
780 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
781 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
782 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
783 ## When the input file's end is reached
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
784 print MAIN $mainSeq . "\n"; # Write the last main DNA sequence to the main sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
785 if ($complement == 1) { # If the complementary file should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
786 $complementSeq = reverse($complementSeq); # Reverse the complementary DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
787 print COMPLEMENT $complementSeq . "\n"; # Write the last complementary DNA sequence to the complementary sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
788 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
789 $sequenceDB{"$sequenceID"} = $mainSeq; # Add sequence to sequence database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
790 $headers{"$sequenceID"} = $header; # Add the header to the header database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
791 push(@sequenceOrder,$sequenceID); # Add this sequence ID to the ordered list of sequences
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
792 $mainSeq = ""; # Empty the main sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
793 $complementSeq = ""; # Empty the complementary sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
794 close (SEQUENCES); # Close the sequence input file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
795 close (COMPLEMENT); # Close the complementary output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
796 close (MAIN); # Close the main output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
797 } else { # If no input file is supplied, then read from stdin instead
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
798 $input = "$tempDir/main.fasta"; # Set up a temporary input file path
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
799 open (MAIN, ">$tempDir/main.fasta"); # Create a temporary file for storing the cleaned sequences representing the main strand
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
800 open (COMPLEMENT, ">$tempDir/complement.fasta"); # Create a temporary file for storing the cleaned sequences representing the complementary strand
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
801 $inputSequenceCount = 0; # Reset input sequence counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
802 while ($sequence = <STDIN>) { # Repeat for every line in the standard input
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
803 chomp($sequence); # Truncate any potential line feeds
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
804 if (substr($sequence,0,1) eq ">") { # If a new FASTA entry is found in the input
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
805 $inputSequenceCount++; # Add one to the input sequence counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
806 print MAIN $mainSeq . "\n"; # Write the previous main DNA sequence to the main sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
807 print MAIN $sequence . " main\n"; # Write the definition line of the new sequence to the main sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
808 $sequenceDB{"$sequenceID"} = $mainSeq; # Add sequence to sequence database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
809 $headers{"$sequenceID"} = $header; # Add the header to the header database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
810 push(@sequenceOrder,$sequenceID); # Add this sequence ID to the ordered list of sequences
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
811 $sequenceID = split(" ",substr($sequence,1));
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
812 $header = $sequence; # Save the sequence header
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
813 $mainSeq = ""; # Empty the main sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
814 if ($complement == 1) { # If the complementary file should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
815 $complementSeq = reverse($complementSeq); # Reverse the complementary DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
816 print COMPLEMENT $complementSeq . "\n"; # Write the previous complementary DNA sequence to the complementary sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
817 print COMPLEMENT $sequence . " complement\n"; # Write the definition line of the new sequence to the complement sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
818 $complementSeq = ""; # Empty the complementary sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
819 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
820 } else { # If this line is just a continuation of the current DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
821 $mseq = $sequence; # Store this part of the DNA sequence in the intermediate varaible $mseq
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
822 $mseq =~ s/[ .-]//g; # Remove any gap characters present in this sequence part (good if input was an alignment)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
823 $mseq =~ tr/[a-z]/[A-Z]/; # Make all letters uppercase
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
824 $mseq =~ s/[^A-Z]//g; # Remove all non-alphabetic characters
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
825 $mseq =~ tr/U/T/; # Exchanges U:s for T:s (Uracil to Thymine, good if input was RNA sequence)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
826 $mainSeq = $mainSeq . $mseq; # Add the intermediate DNA sequence to the end of the main DNA sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
827
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
828 if ($complement == 1) { # If the complementary file should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
829 $cseq = $sequence; # Store this part of the DNA sequence in the intermediate varaible $cseq
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
830 $cseq =~ s/[ .-]//g; # Remove any gap characters present in this sequence part (good if input was an alignment)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
831 $cseq =~ tr/[a-z]/[A-Z]/; # Make all letters uppercase
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
832 $cseq =~ s/[^A-Z]//g; # Remove all non-alphabetic characters
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
833 $cseq =~ tr/ACGTURYSWKMBDHVN/TGCAAYRWSMKVHDBN/; # Replace all characters with its complementary base
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
834 $complementSeq = $complementSeq . $cseq; # Add the intermediate DNA sequence to the end of the complementary DNA sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
835 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
836 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
837 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
838 ## When the input file's end is reached
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
839 print MAIN $mainSeq . "\n"; # Write the last main DNA sequence to the main sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
840 if ($complement == 1) { # If the complementary file should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
841 $complementSeq = reverse($complementSeq); # Reverse the complementary DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
842 print COMPLEMENT $complementSeq . "\n"; # Write the last complementary DNA sequence to the complementary sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
843 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
844 $sequenceDB{"$sequenceID"} = $mainSeq; # Add sequence to sequence database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
845 $headers{"$sequenceID"} = $header; # Add the header to the header database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
846 push(@sequenceOrder,$sequenceID); # Add this sequence ID to the ordered list of sequences
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
847 $mainSeq = ""; # Empty the main sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
848 $complementSeq = ""; # Empty the complementary sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
849 close (COMPLEMENT); # Close the complementary output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
850 close (MAIN); # Close the main output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
851 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
852
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
853 if ($out_sum == 1) { # If summary output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
854 print SUMMARY "Number of sequences in input file: \t$inputSequenceCount\n"; # Write info on the number of input sequences to the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
855 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
856
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
857
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
858 ## Perform HMM-scan
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
859 if ($hmmscan eq "") { # If a pre-computed hmmscan output file is not supplied
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
860 if ($heuristics == 0) { # If HMMER's heuristic filtering should not be used
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
861 $heurMax = "--max"; # Set the heurMax to "--max" (indicating that HMMER should turn off filtering)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
862 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
863 $heurMax = ""; # Set the heurMax to empty (indicating that HMMER should turn on filtering)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
864 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
865 if ($multi_thread == 0) { # If multi-threading is off
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
866 ## Get the current time and output info message
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
867 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
868 if ($silent == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
869 print STDERR "$now : Comparing sequences to HMM database (this may take a long while)...\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
870 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
871 foreach $set (@profileSet) { # Go sequentially through all profile sets to search for
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
872 $hmmPath = $profileDB . "/" . $set . ".hmm"; # Set the path to the HMM-file of the current set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
873 if ($search_eval ne "") { # If E-value cutoff is use for the search
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
874 hmmerSearch("hmmscan --cpu $cpu $heurMax -E $search_eval $hmmPath $tempDir/main.fasta","$tempDir/main.$set.hmmscan","M",$set); # Call HMMER with E-value cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
875 } else { # If score cutoff is use for the search
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
876 hmmerSearch("hmmscan --cpu $cpu $heurMax -T $search_score $hmmPath $tempDir/main.fasta","$tempDir/main.$set.hmmscan","M",$set); # Call HMMER with score cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
877 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
878 if ($complement == 1) { # If the complementary file should be scanned
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
879 if ($search_eval ne "") { # If E-value cutoff is use for the search
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
880 hmmerSearch("hmmscan --cpu $cpu $heurMax -E $search_eval $hmmPath $tempDir/complement.fasta","$tempDir/complement.$set.hmmscan","C",$set); # Call HMMER with E-value cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
881 } else { # If score cutoff is use for the search
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
882 hmmerSearch("hmmscan --cpu $cpu $heurMax -T $search_score $hmmPath $tempDir/complement.fasta","$tempDir/complement.$set.hmmscan","C",$set); # Call HMMER with score cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
883 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
884 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
885 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
886 } else { # If multi-threading is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
887 ## Get the current time and output info message
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
888 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
889 if ($silent == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
890 print STDERR "$now : Doing paralellised comparison to HMM database (this may take a long while)...\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
891 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
892
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
893 ## Determining number of cpus per thread
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
894 if ($complement == 1) { # If the complementary file should be scanned
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
895 $hmmcpu = int(0.5 * $cpu / scalar(@profileSet)); # Assign X CPUs to each thread, X = 0.5 * (TOTAL_CPUs_USED) / (TOTAL_NUMBER_OF_PROFILE_SETS)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
896 } else { # If the complementary file should not be scanned
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
897 $hmmcpu = int($cpu / scalar(@profileSet)); # Assign X CPUs to each thread, X = (TOTAL_CPUs_USED) / (TOTAL_NUMBER_OF_PROFILE_SETS)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
898 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
899 if ($hmmcpu < 1) { # If the number of CPUs per thread is smaller than 1
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
900 $hmmcpu = 1; # Give each thread at least one CPU to work on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
901 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
902 ## Main strand searches...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
903 $cpuCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
904 foreach $set (@profileSet) { # Go through each profile set to investigate
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
905 if ($cpuCount < $cpu) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
906 $cpuCount++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
907 $pid = fork(); # Fork off a copy of this process for this set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
908 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
909 $deceasedPID = wait(); # Wait until a PID is finished, and gather its number
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
910 $pid = fork(); # Fork off a copy of this process for this set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
911 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
912 if ($pid != 0) { # If this is the parent process
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
913 push(@pids,$pid); # Add the new process ID to the list of active process IDs
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
914 } else { # If this is the new child process
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
915 $hmmPath = $profileDB . "/" . $set . ".hmm"; # Set the path to the HMM-file of the current set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
916 if ($search_eval ne "") { # If E-value cutoff is use for the search
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
917 hmmerSearch("hmmscan --cpu $hmmcpu $heurMax -E $search_eval $hmmPath $tempDir/main.fasta","$tempDir/main.$set.hmmscan","M",$set); # Call HMMER with E-value cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
918 } else { # If score cutoff is use for the search
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
919 hmmerSearch("hmmscan --cpu $hmmcpu $heurMax -T $search_score $hmmPath $tempDir/main.fasta","$tempDir/main.$set.hmmscan","M",$set); # Call HMMER with score cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
920 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
921 ## Stop child process...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
922 exit; # Exits the child process
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
923 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
924 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
925 ## Revese strand searches...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
926 if ($complement == 1) { # If the complementary file should be scanned
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
927 foreach $set (@profileSet) { # Go through each profile set to investigate
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
928 if ($cpuCount < $cpu) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
929 $cpuCount++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
930 $pid = fork(); # Fork off a copy of this process for this set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
931 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
932 $deceasedPID = wait(); # Wait until a PID is finished, and gather its number
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
933 $pid = fork(); # Fork off a copy of this process for this set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
934 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
935 if ($pid != 0) { # If this is the parent process
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
936 push(@pids,$pid); # Add the new process ID to the list of active process IDs
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
937 } else { # If this is the new child process
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
938 $hmmPath = $profileDB . "/" . $set . ".hmm"; # Set the path to the HMM-file of the current set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
939 if ($search_eval ne "") { # If E-value cutoff is use for the search
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
940 hmmerSearch("hmmscan --cpu $hmmcpu $heurMax -E $search_eval $hmmPath $tempDir/complement.fasta","$tempDir/complement.$set.hmmscan","C",$set); # Call HMMER with E-value cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
941 } else { # If score cutoff is use for the search
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
942 hmmerSearch("hmmscan --cpu $hmmcpu $heurMax -T $search_score $hmmPath $tempDir/complement.fasta","$tempDir/complement.$set.hmmscan","C",$set); # Call HMMER with score cutoff
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
943 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
944 ## Stop child process...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
945 exit; # Exits the child process
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
946 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
947 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
948 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
949 ## Get the current time and output the active process IDs
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
950 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
951 #print STDERR " $now : Active PIDs: ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
952 #foreach $p (@pids) { # Go through the list of PIDs
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
953 # print STDERR "$p "; # Print the PID
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
954 #}
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
955 #print STDERR "\n"; # Print a new line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
956 do { # Loop until all child PIDs have finished.
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
957 $deceasedPID = wait(); # Wait until a PID is finished, and gather its number
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
958 $now = localtime; # Get the current time
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
959 if ($deceasedPID > -1) { # If the PID that finished wasn't the last active one
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
960 # print STDERR " $now : PID $deceasedPID finished.\n"; # Print finished PID
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
961 } else { # If PID that finished was the last
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
962 print STDERR " $now : All processes finished.\n"; # Print that all PIDs have finished
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
963 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
964 } until (wait() == -1); # Do this loop until all PIDs have finished
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
965 $now = localtime; # Get current time
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
966 print STDERR "$now : Parallel HMM-scan finished.\n"; # Print informative finishing message
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
967 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
968 } else { # If a pre-computed hmmscan file is supplied then
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
969 ## Get the current time and output that the hmmscan step is skipped
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
970 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
971 if ($silent == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
972 print STDERR "$now : Skipping hmmscan! Using $hmmscan as input for the analysis instead.\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
973 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
974 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
975
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
976 ## Analyse HMM-scan output
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
977 ## Get the current time and output info
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
978 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
979 if ($silent == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
980 print STDERR "$now : Analysing results of HMM-scan (this might take quite some time)...\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
981 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
982
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
983 ## Set up output files
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
984 if ($out_table == 1) { # If table output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
985 open (TABLE, ">$output.hmmer.table"); # Create a table output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
986 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
987 if ($out_graph == 1) { # If graphical output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
988 open (GRAPH, ">$output.graph"); # Create a graph output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
989 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
990 if ($out_not == 1) { # If not-found output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
991 open (NOTFOUND, ">$tempDir/$output\_hmmer_no_detections.txt"); # Create a HMMER not-found output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
992 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
993
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
994 $setI = 0; # Set the profile set indicator to zero
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
995 foreach $set (@profileSet) { # Go through all the profile sets to be investigated
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
996 for ($co = 0; $co <= 1; $co++) { # Do main (and complementary) strand analysis in order
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
997 if ($co > 0) { # If main strand analysis is finished
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
998 if ($complement == 1) { # If complementary strand should be analysed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
999 open (HMMOUTPUT, "$tempDir/complement.$set.hmmscan"); # Open hmmscan output for reading
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1000 open (SEQUENCES, "$tempDir/complement.fasta"); # Open complementary sequence file for reading
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1001 if ($out_table == 1) { # If table output is on, write a header for this set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1002 print TABLE "***********************************************************\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1003 print TABLE "$set matches on complementary strand:\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1004 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1005 if ($out_graph == 1) { # If graphical output is on, write a header for this set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1006 print GRAPH "***********************************************************\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1007 print GRAPH "$set matches on complementary strand:\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1008 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1009 } else { # If complementary strand should not be analysed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1010 last; # Exit this loop
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1011 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1012 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1013 if ($co == 0) { # If main strand analysis is not finished
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1014 open (HMMOUTPUT, "$tempDir/main.$set.hmmscan"); # Open hmmscan output for reading
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1015 open (SEQUENCES, "$tempDir/main.fasta"); # Open main sequence file for reading
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1016 if ($out_table == 1) { # If table output is on, write a header for this set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1017 print TABLE "***********************************************************\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1018 print TABLE "$set matches on main strand:\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1019 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1020 if ($out_graph == 1) { # If graphical output is on, write a header for this set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1021 print GRAPH "***********************************************************\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1022 print GRAPH "$set matches on main strand:\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1023 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1024 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1025 ## Read and analyse hmmscan output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1026 while ($line = <HMMOUTPUT>) { # Read in the hmmscan output file, line by line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1027 chomp($line); # Remove any potential line feeds
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1028 if (substr($line,0,13) eq "## New query:") { # If this line begin with "## New query:", then this is a new entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1029 undef %hits; # Empty the hits hash
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1030 undef %evals; # Empty the e-value hash
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1031 undef %scores; # Empty the score hash
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1032 $querytemp = substr($line,14); # Extract everything from this line, except for the start ("## New query:")
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1033 ($query,$length) = split('\t',$querytemp); # Split query name, length and DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1034 if ($co == 0) { # If main strand analysis
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1035 $DNA = $sequenceDB{"$query"};
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1036 } else { # If complementary strand
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1037 $cseq = $sequenceDB{"$query"}; # Store the DNA sequence in the intermediate varaible $cseq
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1038 $cseq =~ tr/[a-z]/[A-Z]/; # Make all letters uppercase
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1039 $cseq =~ s/[^A-Z]//g; # Remove all non-alphabetic characters
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1040 $cseq =~ tr/ACGTURYSWKMBDHVN/TGCAAYRWSMKVHDBN/; # Replace all characters with its complementary base
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1041 $DNA = reverse($cseq);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1042 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1043 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1044
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1045 if ($line ne "//") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1046 ## Find domain annotations...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1047 ($query,$matchProfile,$length,$domNo,$sign,$score,$bias,$cE,$iE,$hmmFrom,$hmmTo,$hmmends,$queryFrom,$queryTo,$queryends,$envFrom,$envTo,$envends,$acc) = split('\t',$line); # Split the line into a collection of stat variables
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1048
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1049 $useQueryFrom = $queryFrom;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1050 if ($hmmFrom > 1) { # If the HMM-profile is not matched from the beginning
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1051 $hmmDiff = $hmmFrom - 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1052 if ($useQueryFrom > $hmmDiff) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1053 $useQueryFrom = $queryFrom - $hmmDiff;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1054 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1055 $useQueryFrom = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1056 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1057 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1058
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1059 if (uc($anchor) eq "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1060 $anchorLen = $hmmTo - $hmmFrom + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1061 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1062 $anchorLen = $anchor;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1063 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1064
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1065
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1066 $query_profile_match = $query . ":" . $matchProfile;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1067 $profileExists = 0; # Assume that the newly found match profile ($matchProfile) is not already found for this sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1068 if (exists($hits{$query_profile_match})) { # If a profile from the list is the same as the match profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1069 ($hitFrom,$hitTo,$hitProfile,$hitScore,$hitE) = split('\t',$hits{$query_profile_match}); # Split the entry in list into stat variables
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1070 if ($iE < $hitE) { # If the new match profile has a smaller E-value than the one from the list
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1071 $hits{$query_profile_match} = "$useQueryFrom\t$envTo\t$matchProfile\t$score\t$iE\t$anchorLen"; # Replace the data in the hit list with the data for the newly found match profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1072 $evals{$query_profile_match} = $iE; # Replace the E-value in the hit list with the E-value for the newly found match profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1073 $scores{$query_profile_match} = $score; # Replace the score in the hit list with the score for the newly found match profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1074 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1075 $profileExists = 1; # Indicate that this match profile was found in the hit list
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1076 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1077 if ($profileExists == 0) { # If the match profile was not found in the hit list
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1078 if (($iE <= $E) && ($score >= $S)) { # If this hits lives up to the minimal score and E-value cutoffs
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1079 $hits{$query_profile_match} = "$useQueryFrom\t$envTo\t$matchProfile\t$score\t$iE\t$anchorLen"; # Add the data for the newly found match profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1080 $evals{$query_profile_match} = $iE; # Add the E-value for the newly found match profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1081 $scores{$query_profile_match} = $score; # Add the score for the newly found match profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1082 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1083 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1084 } else { # If the line only contains "//", the end of this sequence's hmmscan entry is reached
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1085 ## Save analysis results
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1086 @sortedKeys = sort {$hits{$a} <=> $hits{$b}} keys(%hits); # Sort the the list of hits numerically ascending (smallest first)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1087 undef @sortedHits;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1088 undef @scores;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1089 undef @evals;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1090 foreach $key (@sortedKeys) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1091 push(@sortedHits, $hits{$key}); # Add the hit to the list of hits numerically ascending (smallest first)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1092 push(@scores, $scores{$key}); # Add the score to the scores array
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1093 push(@evals, $evals{$key}); # Add the E-value to the evals array
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1094 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1095
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1096 ## If the number of hits > N, the min eval < E and the max score > S then include query sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1097 ## OR if a single domain satisfies the thresholds and this is allowed, include it!
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1098 if ( ((scalar(@sortedHits) >= $N) && (min(@evals) <= $E) && (max(@scores) >= $S)) ||
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1099 ((scalar(@sortedHits) > 0) && ($allow_single_E >= 0) && (min(@evals) <= $allow_single_E) && (max(@scores) >= $allow_single_score)) ) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1100 if ($debug == 1) { # If debugging mode is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1101 print STDERR $query . " :\t" . scalar(@sortedHits) . "\t" . min(@evals) . "\t" . max(@scores) . "\n"; # Print some top hit statistics
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1102 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1103
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1104 ## Save some total stats to be able to determine origin of ITS sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1105 if (scalar(@evals) > 0) { # If there are any E-values stored
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1106 $averageE = sum(@evals) / scalar(@evals); # Calculate the average E-value for this profile set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1107 $averageScore = sum(@scores) / scalar(@scores); # Calculate the average score for this profile set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1108 $numberOfDomains = scalar(@sortedHits); # Calculate the number of domains matched on this sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1109
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1110 #$scoreSum = sum(@scores) / @modelCount[$setI]; # Calculate score sum as: sum / (no. of profiles of this given type)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1111 $scoreSum = sum(@scores) / 4; # Calculate score sum as: sum / (no. of profiles of this given type)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1112
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1113 $saveThis = "$query\t$set\t$co\t$numberOfDomains\t$averageE\t$averageScore\t$scoreSum\t$DNA\t"; # Collect the variables to save for this sequence and this profile set
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1114 foreach $hit (@sortedHits) { # Go through the list of hits and add specific information to save from each hit
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1115 ($hitFrom,$hitTo,$hitProfile,$hitScore,$hitE,$hitanchorlen) = split('\t',$hit); # Extract information from this hit
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1116 $saveThis = $saveThis . "$hitFrom;$hitTo;$hitProfile;$hitScore;$hitE;$hitanchorlen\t"; # Add information to the list of variables to save
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1117 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1118 #push(@allHits, $saveThis); # Add this information to the collection of all hits for this sequence, across all profile sets
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1119 if (exists($allHits{$query})) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1120 $allHits{$query} = $allHits{$query} . "\n" . $saveThis;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1121 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1122 $allHits{$query} = $saveThis;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1123 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1124 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1125
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1126 if ($out_table == 1) { # If table output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1127 print TABLE $query . "\t" . $length . "\t"; # Print query and length information to table
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1128 foreach $hit (@sortedHits) { # Go through each hit in the hit list
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1129 ($hitFrom,$hitTo,$hitProfile,$hitScore,$hitE,$hitanchorlen) = split('\t',$hit); # Extract data corresponding to this hit
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1130 print TABLE "$hitFrom - $hitTo: $hitProfile ($hitScore, $hitE)\t"; # Print hit information to table
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1131 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1132 print TABLE "\n"; # Print new line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1133 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1134
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1135 if ($out_graph == 1) { # If graphical output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1136 print GRAPH ">> " . $query . "\t" . $length . " bp\n"; # Print a sequence header
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1137 $insertPoint = 0; # Set the domain insert point to beginning of line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1138 $hi = 0; # Set hit number to zero
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1139 foreach $hit (@sortedHits) { # Go through the hit list
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1140 ($hitFrom,$hitTo,$hitProfile,$hitScore,$hitE,$anchorLen) = split('\t',$hit); # Split the hit into stat variables
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1141 if ($graph_scale == 0) { # If the graph scale is scaled individually to 100% for each sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1142 $pFrom = $hitFrom / $length * 100; # Set the profile start on graph relative to its position in the sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1143 $pTo = $hitTo / $length * 100; # Set the profile end on graph relative to its position in the sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1144 $pEnd = 100; # Set the end of the sequence graph to be at 100
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1145 } else { # If the scale is the same for all sequences
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1146 $pFrom = $hitFrom * $graph_scale; # Set the profile start on graph scaled to the parameter given
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1147 $pTo = $hitTo * $graph_scale; # Set the profile end on graph scaled to the parameter given
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1148 $pEnd = $length * $graph_scale; # Set the end of the sequence graph to be at the end of the sequence scaled to the parameter given
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1149 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1150 for ($insertPoint = $insertPoint; $insertPoint <= $pFrom; $insertPoint++) { # Go forward through the sequence, moving the insert point one step at a time until the beginning of the next profile is reached
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1151 print GRAPH "-"; # Print a "-"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1152 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1153 print GRAPH substr($hitProfile,2,3); # When the profile is reached, print its name
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1154 $insertPoint = $insertPoint + 3; # Move the insert point three steps forward, to account for the inserted name
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1155 ($nextHitStart,$nextHitEnd,$nextProfile) = split('\t',@sortedHits[$hi + 1]); # Check where the next hit in the list is located
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1156 if (($nextHitStart <= $hitTo) && ($nextHitStart > 0)) { # If the next hit in the list overlaps with this profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1157 if ($graph_scale == 0) { # If the scale is relative
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1158 $pTo = $nextHitStart / $length * 100 - 1; # Change the profile end on the graph to be where this next profile starts
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1159 } else { # If the scale is the same for all sequences
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1160 $pTo = $nextHitStart * $graph_scale - 1; # Change the profile end on the graph to be where this next profile starts
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1161 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1162 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1163 for ($insertPoint = $insertPoint; $insertPoint <= $pTo; $insertPoint++) { # Go forward through the sequence, moving the insert point one step at a time until the end of the current profile is reached
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1164 print GRAPH "="; # Print a "="
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1165 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1166 if (($nextHitStart <= $hitTo) && ($nextHitStart > 0)) { # If the next hit in the list overlaps with this profile
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1167 print GRAPH ">"; # Print a ">" to indicate the profile overlap
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1168 $insertPoint++; # Move the insert point one additional step forward to account for the ">" inserted
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1169 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1170 $hi++; # Increase the hit number by one
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1171 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1172 for ($insertPoint = $insertPoint; $insertPoint <= $pEnd; $insertPoint++) { # If there is no more profile matches to sequence, go forward through the sequence, moving the insert point one step at a time until the end of the sequence is reached
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1173 print GRAPH "-"; # Print a "-"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1174 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1175 print GRAPH "\n"; # Print a new line, indicating the end of this sequence entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1176 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1177 } else { # If this sequence didn't find any good-enough profile matches
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1178 if ($out_not == 1) { # If not-found output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1179 print NOTFOUND $query . "\n"; # Print the name of this query to the not-found list
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1180 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1181 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1182 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1183 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1184 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1185 close (SEQUENCES); # Close the input sequence file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1186 close (HMMOUTPUT); # Close the hmmscan output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1187 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1188 $setI++; # Add one to the profile set indicator
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1189 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1190
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1191 ## Close output files
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1192 if ($out_table == 1) { # If table output is on, close the table file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1193 close (TABLE);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1194 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1195 if ($out_graph == 1) { # If graphical output is on, close the graph file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1196 close (GRAPH);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1197 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1198 if ($out_not == 1) { # If not-found output is on, close the not-found file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1199 close (NOTFOUND);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1200 #$profileCount = scalar(@profileSet); # Count the number of profile sets
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1201 #if ($complement == 1) { # If complementary strand was scanned
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1202 # $profileCount = $profileCount * 2; # Double the number of profile sets that was investigated (and thus the number of not-founds that could at max be found)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1203 #}
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1204 #`sort $tempDir/hmmer_no_detections.txt | uniq -c | grep " *$profileCount " | sed "s/ *$profileCount //" > $output\_no_detections.txt`; # Sort the not-found list, count the number of profile sets having no matches for each query. Save those that have only non-matches to the hmmer-not-found file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1205 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1206
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1207
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1208 ## Create total collected output and FASTA output
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1209
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1210 if ($out_results == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1211 open (RESULTS, ">$output.extraction.results"); # Create a results file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1212 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1213 open (RAWOUT, ">$tempDir/ITSx_output.raw"); # Create a raw output file for ALL data
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1214 open (PROBLEM, ">$output.problematic.txt"); # Create a file for problematic entries
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1215 $foundProblem = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1216 if ($out_pos == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1217 open (POS, ">$output.positions.txt"); # Create a positions file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1218 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1219 if ($out_fasta == 1) { # If FASTA output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1220 open (FASTA, ">$output.full.fasta"); # Create a FASTA output file for found sequences
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1221 if ($allow_reorder == 0) { # If reordering of domains is not allowed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1222 open (CHIMERA, ">$output.chimeric.fasta"); # Create a FASTA file for potential chimera sequences with profile matches in the wrong order
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1223 $foundChimera = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1224 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1225 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1226 open (FULLPARTIAL, ">$output.full_and_partial.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1227 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1228 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1229
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1230 if ($out_joined == 1) { # If SSU FASTA output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1231 open (JOINED, ">$output.joined.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1232 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1233 if ($out_ssu == 1) { # If SSU FASTA output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1234 open (SSU, ">$output.SSU.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1235 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1236 if ($out_lsu == 1) { # If LSU FASTA output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1237 open (LSU, ">$output.LSU.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1238 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1239 if ($out_58S == 1) { # If 5.8S FASTA output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1240 open (MID, ">$output.5_8S.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1241 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1242 if ($out_its1 == 1) { # If ITS1 FASTA output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1243 open (ITS1, ">$output.ITS1.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1244 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1245 open (ITS1PARTIAL, ">$output.ITS1.full_and_partial.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1246 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1247 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1248 if ($out_its2 == 1) { # If ITS2 FASTA output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1249 open (ITS2, ">$output.ITS2.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1250 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1251 open (ITS2PARTIAL, ">$output.ITS2.full_and_partial.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1252 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1253 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1254 if ($out_concat == 1) { # If concatenated output should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1255 open (CONCAT, ">$output.concat.fasta");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1256 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1257
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1258 undef @sortedHits; # Empty the array of sorted hits
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1259
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1260 # @sortedHits = sort @allHits; # Sort the full list of hits in alphabetical order (to be able to analyse all sequences with same ID at once)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1261 if ($out_not == 1) { # If not-found output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1262 open (NOTFOUND, ">$output\_no_detections.txt"); # Create a not-found output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1263 $noDetect = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1264 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1265 foreach $sequenceID (@sequenceOrder) { # Sort the full list of hits in their original order (to be able to analyse all sequences with same ID at once)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1266 if ($sequenceID ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1267 $countsInList = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1268 if (exists($allHits{$sequenceID})) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1269 @allHits = split('\n',$allHits{$sequenceID});
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1270 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1271 undef @allHits;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1272 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1273 foreach $line (@allHits) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1274 @item = split('\t',$line); # Split the line into an array
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1275 if (@item[0] eq $sequenceID) { # If this item corresponds to the current sequence ID
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1276 push(@sortedHits,$line); # Add it to the sorted list of hits
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1277 $countsInList++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1278 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1279 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1280 if ($countsInList == 0) { # If no matches were found
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1281 if ($out_not == 1) { # If not-found output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1282 print NOTFOUND "$sequenceID\n"; # Output the sequence ID
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1283 $noDetect++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1284 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1285 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1286 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1287 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1288 if ($out_not == 1) { # If not-found output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1289 close (NOTFOUND);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1290 if ($noDetect == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1291 `rm $output\_no_detections.txt 2> /dev/null`;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1292 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1293 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1294
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1295 ## Set all counts for different ITS types to zero
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1296 undef @itsCounts;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1297 $itsChimeric = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1298 $itsMain = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1299 $itsCompl = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1300
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1301 push(@sortedHits,"--END--"); # Add a last item to the sorted list, so that all items are securely saved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1302 $lc = 1; # Set the line count to one
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1303 foreach $line (@sortedHits) { # Go through the list of found hits
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1304 print RAWOUT "$line\n"; # Write the raw data associated with this hit to the raw data output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1305 @item = split('\t',$line); # Split the line into an array
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1306 ## If this sequence ID is the same as the saved ones, then add it, else empty the array sequence ID and save
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1307 if ((@seqID[0] ne @item[0]) && (@item[0] ne "") || ($lc > scalar(@sortedHits))) { # If this sequence ID is not the same as the last one and is non-empty, or if the end of the list has been reached
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1308 ## Save profile-type which is most likely...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1309 if ($priority eq "sum") { # If the sum-of-scores algorithm should be used to determine the most likely profile-type
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1310 ## Reset variables to unrealisticly high or low values
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1311 $best = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1312 $bestCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1313 $bestEval = 1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1314 $bestScore = -1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1315 $bestSum = -1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1316
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1317 for ($i = 0; $i < scalar(@seqScoreSum); $i++) { # Go through all sum-of-scores entries
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1318 if (@seqScoreSum[$i] > $bestSum) { # If the current value is larger than the previous top value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1319 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1320 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1321 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1322 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1323 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1324 $bestSum = @seqScoreSum[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1325 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1326 if (@seqScoreSum[$i] == $bestSum) { # If the current value is equal to the previous top value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1327 if (@seqDomCounts[$i] > $bestCount) { # If the current domain count is larger than the previous top domain count
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1328 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1329 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1330 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1331 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1332 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1333 $bestSum = @seqScoreSum[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1334 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1335 if (@seqDomCounts[$i] == $bestCount) { # If the current domain count is equal to the previous top domain count
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1336 if (@seqAvgE[$i] < $bestEval) { # If the current E-value is smaller than the previous top E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1337 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1338 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1339 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1340 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1341 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1342 $bestSum = @seqScoreSum[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1343 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1344 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1345 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1346 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1347 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1348 if ($priority eq "domains") { # If the number of found domains should be used to determine the most likely profile-type
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1349 ## Reset variables to unrealisticly high or low values
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1350 $best = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1351 $bestCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1352 $bestEval = 1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1353 $bestScore = -1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1354
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1355 for ($i = 0; $i < scalar(@seqDomCounts); $i++) { # Go through all domain count entries
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1356 if (@seqDomCounts[$i] > $bestCount) { # If the current domain count is larger than the previous top domain count
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1357 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1358 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1359 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1360 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1361 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1362 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1363 if (@seqDomCounts[$i] == $bestCount) { # If the current domain count is equal to the previous top domain count
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1364 if (@seqAvgE[$i] < $bestEval) { # If the current E-value is smaller than the previous top E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1365 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1366 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1367 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1368 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1369 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1370 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1371 if (@seqAvgE[$i] == $bestEval) { # If the current E-value is equal to the previous top E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1372 if (@seqAvgScore[$i] > $bestScore) { # If the current average score is larger than the previous top average score
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1373 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1374 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1375 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1376 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1377 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1378 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1379 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1380 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1381 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1382 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1383 if ($priority eq "eval") { # If the average E-value should be used to determine the most likely profile-type
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1384 ## Reset variables to unrealisticly high or low values
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1385 $best = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1386 $bestCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1387 $bestEval = 1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1388 $bestScore = -1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1389
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1390 for ($i = 0; $i < scalar(@seqDomCounts); $i++) { # Go through all domain counts entries
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1391 if (@seqAvgE[$i] < $bestEval) { # If the current E-value is smaller than the previous top E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1392 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1393 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1394 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1395 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1396 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1397 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1398 if (@seqAvgE[$i] == $bestEval) { # If the current E-value is equal to the previous top E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1399 if (@seqAvgScore[$i] > $bestScore) { # If the current average score is larger than the previous top average score
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1400 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1401 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1402 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1403 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1404 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1405 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1406 if (@seqAvgScore[$i] == $bestScore) { # If the current average score is equal to the previous top average score
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1407 if (@seqDomCounts[$i] > $bestCount) { # If the current number of domains is larger than the previous top number of domains
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1408 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1409 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1410 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1411 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1412 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1413 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1414 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1415 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1416 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1417 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1418 if ($priority eq "score") { # If the average score should be used to determine the most likely profile-type
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1419 ## Reset variables to unrealisticly high or low values
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1420 $best = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1421 $bestCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1422 $bestEval = 1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1423 $bestScore = -1000;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1424
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1425 for ($i = 0; $i < scalar(@seqDomCounts); $i++) { # Go through all domain counts entries
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1426 if (@seqAvgScore[$i] > $bestScore) { # If the current average score is larger than the previous top average score
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1427 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1428 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1429 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1430 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1431 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1432 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1433 if (@seqAvgScore[$i] == $bestScore) { # If the current average score is equal to the previous top average score
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1434 if (@seqAvgE[$i] < $bestEval) { # If the current E-value is smaller than the previous top E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1435 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1436 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1437 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1438 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1439 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1440 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1441 if (@seqAvgE[$i] == $bestEval) { # If the current E-value is equal to the previous top E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1442 if (@seqDomCounts[$i] > $bestCount) { # If the current number of domains is larger than the previous top number of domains
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1443 $best = $i; # Set the best value to be the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1444 ## Set all other best variables to those corresponding to the current value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1445 $bestCount = @seqDomCounts[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1446 $bestEval = @seqAvgE[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1447 $bestScore = @seqAvgScore[$i];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1448 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1449 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1450 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1451 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1452 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1453
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1454 if (@seqID[$best] ne "") { # If the sequence ID of the most likely profile is not empty
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1455
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1456 $allanchorLens = @anchorLens[$best]; # Get the best anchor lengths
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1457 @allanchorLens = split(',', $allanchorLens); # Split the anchor lens into an array
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1458
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1459 $chimeric = 0; # Assume the sequence is not chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1460 if ($allow_reorder == 0) { # If re-order of domain is not allowed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1461 $domain_order = @allSeqDomains[$best]; # Gather the order the domains are found in
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1462 @domain_order = split(' ',$domain_order); # Split the list into an array
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1463 @sorted_domain_order = sort {$a cmp $b} @domain_order; # Sort the array alphabetically
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1464 for ($di = 0; $di <= scalar(@domain_order); $di++) { # Go through the sorted array
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1465 if ((@domain_order[$di] ne @sorted_domain_order[$di]) || (@problemCode[$best] =~ m/C/)) { # Check if the order of the arrays differ at any poiny
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1466 $chimeric = 1; # If they do differ, mark the sequence as chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1467 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1468 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1469 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1470
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1471 $seqDNALength = length(@seqDNA[$best]); # Get the length of the DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1472
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1473 ## Print sequence and match data...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1474 ## Order of columns in the output file:
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1475 ## ID Length Type Main/Compl Domains Avg.Eval Avg.Score Start End Start_domain End_domain Chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1476 if ($out_results == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1477 print RESULTS @seqID[$best] . "\t" . $seqDNALength . "\t" . @seqITSType[$best] . "\t" . @seqCompl[$best] . "\t" . @seqDomCounts[$best] . "\t" . @seqAvgE[$best] . "\t" . @seqAvgScore[$best] . "\t" . @seqScoreSum[$best] . "\t" . @dnaStart[$best] . "\t" . @dnaEnd[$best] . "\t" . @startDomain[$best] . "\t" . @endDomain[$best] . "\t"; # Print sequence and match data to the results file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1478 if ($chimeric == 1) { # If the sequence was regarded chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1479 print RESULTS "Chimeric\t"; # Add a chimeric tag to the entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1480 } else { # If not chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1481 print RESULTS "\t"; # Add an empty column
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1482 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1483 $allDomains = @allSeqDomains[$best]; # Get the domain order of the entry
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1484 $allDomains =~ tr/ /,/; # Replace spaces with commas in the domain order string
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1485 $allDomains = substr($allDomains,0,length($allDomains) - 1); # Remove the last character (a comma)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1486 print RESULTS $allDomains; # Write the domain order to the results file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1487 print RESULTS "\t"; # Write a tab to the results file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1488 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1489
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1490 if ($out_pos == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1491 $out_all_pos = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1492 if ($out_all_pos == 1) { # Output positions of all domains
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1493 $seqPartLen = @dnaEnd[$best] - @dnaStart[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1494 if ($seqPartLen < 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1495 $seqPartLen = $seqPartLen * -1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1496 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1497 ## Print the positions of all identified domains to the position file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1498 print POS @seqID[$best] . "\t" . $seqDNALength . " bp." . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1499 if (@problemCode[$best] !~ m/S/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1500 print POS "SSU: " . @ssuStart[$best] . "-" . @ssuEnd[$best] . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1501 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1502 print POS "SSU: Not found\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1503 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1504 if ((@problemCode[$best] =~ m/X/) || ((@problemCode[$best] =~ m/[15]/) && (@problemCode[$best] =~ m/S/))) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1505 print POS "ITS1: Not found\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1506 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1507 if (@problemCode[$best] =~ m/O/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1508 print POS "ITS1: " . (@ssuEnd[$best] + 1) . "-" . (@ssuEnd[$best] + 1) . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1509 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1510 print POS "ITS1: " . @its1Start[$best] . "-" . @its1End[$best] . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1511 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1512 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1513 if (@problemCode[$best] !~ m/[125OP]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1514 print POS "5.8S: " . @midStart[$best] . "-" . @midEnd[$best] . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1515 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1516 if (@problemCode[$best] =~ m/5/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1517 print POS "5.8S: Not found\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1518 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1519 if (@problemCode[$best] =~ m/1/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1520 print POS "5.8S: No start\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1521 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1522 if (@problemCode[$best] =~ m/O/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1523 print POS "5.8S: Overlap SSU\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1524 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1525 if (@problemCode[$best] =~ m/P/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1526 print POS "5.8S: Overlap LSU\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1527 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1528 print POS "5.8S: No end\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1529 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1530 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1531 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1532 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1533 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1534 if ((@problemCode[$best] =~ m/Y/) ||((@problemCode[$best] =~ m/[25]/) && (@problemCode[$best] =~ m/L/))) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1535 print POS "ITS2: Not found\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1536 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1537 if (@problemCode[$best] =~ m/O/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1538 print POS "ITS2: " . (@lsuStart[$best] - 1) . "-" . (@lsuStart[$best] - 1) . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1539 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1540 print POS "ITS2: " . @its2Start[$best] . "-" . @its2End[$best] . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1541 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1542 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1543 if (@problemCode[$best] !~ m/L/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1544 print POS "LSU: " . @lsuStart[$best] . "-" . @lsuEnd[$best] . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1545 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1546 print POS "LSU: Not found" . "\t";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1547 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1548
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1549 if (@problemCode[$best] =~ m/5/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1550 print POS "Broken or partial sequence, no 5.8S! ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1551 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1552 if (@problemCode[$best] =~ m/[12]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1553 print POS "Broken or partial sequence, only partial 5.8S! ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1554 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1555 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1556 if (@problemCode[$best] =~ m/B/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1557 print POS "ITS region too long! ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1558 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1559 if (@problemCode[$best] =~ m/O/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1560 print POS "5.8S seem to overlap with SSU! ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1561 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1562 if (@problemCode[$best] =~ m/P/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1563 print POS "5.8S seem to overlap with LSU! ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1564 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1565 if (@problemCode[$best] =~ m/C/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1566 print POS "Chimeric! ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1567 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1568 print POS "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1569
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1570 } else { # Output only ITS positions
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1571 $seqPartLen = @dnaEnd[$best] - @dnaStart[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1572 print POS @seqID[$best] . "\t" . $seqDNALength . " bp." . "\t" . "ITS1: " . @its1Start[$best] . "-" . @its1End[$best] . "\t" . "ITS2: " . @its2Start[$best] . "-" . @its2End[$best] . "\n"; # Print the positions of the ITS sequences to the position file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1573 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1574 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1575
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1576 if (@problem[$best] ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1577 $foundProblem++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1578 print PROBLEM @seqID[$best] . "\t" . @problem[$best] . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1579 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1580
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1581 ## Set extended type string (the string going into the definition line of the FASTA file)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1582 $extendedType = $profileIndex{@seqITSType[$best]} . " ITS sequence";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1583 @itsCounts[ord(@seqITSType[$best])]++; # Add one ITS to the appropriate counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1584
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1585 if (@seqCompl[$best] == 1) { # If domains were found on complementary strand
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1586 $extendedStrand = "complementary strand"; # Set the strand string to complementary
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1587 $itsCompl++; # Add one to the complementary strand counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1588 } else { # If domains were found on the main strand
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1589 $extendedStrand = "main strand"; # Set the strand string to main
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1590 $itsMain++; # Add one to the main strand counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1591 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1592
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1593 ## Print (extracted) ITS sequence...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1594 if ($out_fasta == 1) { # If FASTA-output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1595 if ($truncate == 0) { # If the whole sequence should be kept in output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1596 if ($chimeric == 0) { # If the sequence is not chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1597 if (@problemCode[$best] !~ m/[SL]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1598 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1599 print FASTA $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1600 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1601 print FASTA ">" . @seqID[$best] . "|" . @seqITSType[$best] ." " . $extendedType . " (" . $seqDNALength . " bp) on " . $extendedStrand . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1602 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1603 print FASTA @seqDNA[$best] . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1604 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1605 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1606 if (@problemCode[$best] !~ m/[SL125]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1607 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1608 print FULLPARTIAL $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1609 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1610 print FULLPARTIAL ">" . @seqID[$best] . "|" . @seqITSType[$best] ." " . $extendedType . " (" . $seqDNALength . " bp) Full ITS region on " . $extendedStrand . "\n"; # Write FASTA defline
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1611 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1612 print FULLPARTIAL @seqDNA[$best] . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1613 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1614 $its1PartLen = @its1End[$best] - @its1Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1615 $its2PartLen = @its2End[$best] - @its2Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1616 if (($out_partial < $its1PartLen) && ($out_partial < $its2PartLen)) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1617 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1618 print FULLPARTIAL $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1619 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1620 print FULLPARTIAL ">" . @seqID[$best] . "|" . @seqITSType[$best] ." " . $extendedType . " (" . $seqDNALength . " bp) Partial ITS region on " . $extendedStrand . "\n"; # Write FASTA defline
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1621 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1622 print FULLPARTIAL @seqDNA[$best] . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1623 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1624 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1625 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1626 } else { # If sequence is regarded chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1627 $itsChimeric++; # Add one to the chimeric counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1628 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1629 print CHIMERA $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1630 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1631 print CHIMERA ">" . @seqID[$best] . "|" . @seqITSType[$best] ." Chimeric " . $extendedType . " (" . length(@seqDNA[$best]) . " bp) on " . $extendedStrand . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1632 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1633 print CHIMERA @seqDNA[$best] . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1634 $foundChimera++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1635 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1636 } else { # If only the ITS part of the sequence should be saved to output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1637 $fastaStartPoint = @dnaStart[$best] - 1; # Start extraction at the start of the first domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1638 $fastaEndPoint = @dnaEnd[$best]+10; # End extraction 10 bp after the last domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1639
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1640 if (@seqDomCounts[$best] > 1) { # If more than one domain was found
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1641 if (substr(@startDomain[$best],0,5) eq "1_SSU") { # If the first domain was SSU
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1642 $fastaStartPoint = @ssuEnd[$best]; # Set the start point of the extraction to the end of the SSU domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1643 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1644 if (substr(@startDomain[$best],0,5) eq "4_LSU") { # If the first domain was LSU
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1645 $fastaStartPoint = @lsuEnd[$best]; # Set the start point of the extraction to the end of the LSU domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1646 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1647 if (substr(@endDomain[$best],0,5) eq "1_SSU") { # If the last domain was SSU
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1648 $fastaEndPoint = @ssuStart[$best] - 1; # Set the end point of the extraction to the start of the SSU domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1649 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1650 if (substr(@endDomain[$best],0,5) eq "4_LSU") { # If the last domain was LSU
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1651 $fastaEndPoint = @lsuStart[$best] - 1; # Set the end point of the extraction to the start of the LSU domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1652 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1653 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1654
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1655 if ($fastaStartPoint < 0) { # If the start point is smaller than zero, set the start point to zero
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1656 $fastaStartPoint = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1657 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1658 if ($fastaEndPoint > length(@seqDNA[$best])) { # If the end point is larger than the length of the sequence, set the end point to the sequence end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1659 $fastaEndPoint = length(@seqDNA[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1660 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1661
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1662 $fastaLength = $fastaEndPoint - $fastaStartPoint + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1663
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1664 if ($chimeric == 0) { # If the sequence is not chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1665 if (@problemCode[$best] !~ m/[SL]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1666 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1667 print FASTA $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1668 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1669 print FASTA ">" . @seqID[$best] . "|" . @seqITSType[$best] . " " . $extendedType . " (" . $fastaLength . " bp) on " . $extendedStrand . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1670 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1671 if ($anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1672 if ($fastaStartPoint - @allanchorLens[0] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1673 print FASTA substr(@seqDNA[$best],$fastaStartPoint - @allanchorLens[0],$fastaLength + @allanchorLens[0] + @allanchorLens[3]) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1674 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1675 print FASTA substr(@seqDNA[$best],0,$fastaLength + $fastaStartPoint + @allanchorLens[3]) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1676 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1677 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1678 print FASTA substr(@seqDNA[$best],$fastaStartPoint,$fastaLength) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1679 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1680 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1681 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1682 if (@problemCode[$best] !~ m/[SL125]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1683 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1684 print FULLPARTIAL $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1685 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1686 print FULLPARTIAL ">" . @seqID[$best] . "|" . @seqITSType[$best] ." " . $extendedType . " (" . $seqDNALength . " bp) Full ITS region on " . $extendedStrand . "\n"; # Write FASTA defline
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1687 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1688 print FULLPARTIAL substr(@seqDNA[$best],$fastaStartPoint,$fastaLength) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1689 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1690 $its1PartLen = @its1End[$best] - @its1Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1691 $its2PartLen = @its2End[$best] - @its2Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1692 if (($out_partial < $its1PartLen) && ($out_partial < $its2PartLen)) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1693 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1694 print FULLPARTIAL $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1695 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1696 print FULLPARTIAL ">" . @seqID[$best] . "|" . @seqITSType[$best] ." " . $extendedType . " (" . $seqDNALength . " bp) Partial ITS region on " . $extendedStrand . "\n"; # Write FASTA defline
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1697 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1698 print FULLPARTIAL substr(@seqDNA[$best],$fastaStartPoint,$fastaLength) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1699 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1700 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1701 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1702 } else { # If sequence is regarded chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1703 $itsChimeric++; # Add one to the chimeric counter
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1704 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1705 print CHIMERA $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1706 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1707 print CHIMERA ">" . @seqID[$best] . "|" . @seqITSType[$best] . " Chimeric " . $extendedType . " (" . $fastaLength . " bp) on " . $extendedStrand . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1708 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1709 print CHIMERA @seqDNA[$best] . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1710 $foundChimera++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1711 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1712 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1713 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1714
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1715 if ($out_joined == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1716 if ($chimeric == 0) { # If the sequence is not chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1717 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1718 print JOINED $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1719 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1720 print JOINED ">" . @seqID[$best] . "|" . @seqITSType[$best] . " " . $extendedType . " (" . $fastaLength . " bp) From domain " . @startDomain[$best] . " to " . @endDomain[$best] . " on " . $extendedStrand . " Found domains: "; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1721 print JOINED substr(@allSeqDomains[$best],0,length(@allSeqDomains[$best]) - 1) . "\n"; # Write domain order
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1722 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1723 print JOINED substr(@seqDNA[$best],$fastaStartPoint,$fastaLength) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1724 } else { # If sequence is regarded chimeric
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1725 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1726 print JOINED $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1727 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1728 print JOINED ">" . @seqID[$best] . "|" . @seqITSType[$best] . " Chimeric " . $extendedType . " (" . $fastaLength . " bp) From domain " . @startDomain[$best] . " to " . @endDomain[$best] . " on " . $extendedStrand . " Found domains: "; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1729 print JOINED substr(@allSeqDomains[$best],0,length(@allSeqDomains[$best]) - 1) . "\n"; # Write domain order
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1730 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1731 print JOINED substr(@seqDNA[$best],$fastaStartPoint,$fastaLength) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1732 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1733 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1734
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1735 ## Write SSU sequence to file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1736 if ($out_ssu == 1) { # If SSU output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1737 if (@problemCode[$best] !~ m/S/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1738 if ($only_full == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1739 $seqPartLen = @ssuEnd[$best] - @ssuStart[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1740 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1741 print SSU $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1742 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1743 print SSU ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|SSU " . "Extracted SSU sequence (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1744 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1745 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1746 if (@ssuStart[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1747 print SSU substr(@seqDNA[$best], @ssuStart[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1748 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1749 print SSU substr(@seqDNA[$best], 0, @ssuStart[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1750 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1751 print SSU substr(@seqDNA[$best], @ssuStart[$best] - 1, @ssuEnd[$best] - @ssuStart[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1752 if (@ssuEnd[$best] - @ssuStart[$best] + 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1753 print SSU substr(@seqDNA[$best], @ssuEnd[$best], $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1754 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1755 print SSU substr(@seqDNA[$best], @ssuEnd[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1756 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1757 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1758 print SSU substr(@seqDNA[$best], @ssuStart[$best] - 1, @ssuEnd[$best] - @ssuStart[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1759 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1760 print SSU "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1761 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1762 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1763 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1764
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1765 ## Write LSU sequence to file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1766 if ($out_lsu == 1) { # If LSU output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1767 if ($only_full == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1768 if (@problemCode[$best] !~ m/L/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1769 $seqPartLen = @lsuEnd[$best] - @lsuStart[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1770 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1771 print LSU $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1772 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1773 print LSU ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|LSU " . "Extracted LSU sequence (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1774 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1775 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1776 if (@lsuStart[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1777 print LSU substr(@seqDNA[$best], @lsuStart[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1778 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1779 print LSU substr(@seqDNA[$best], 0, @lsuStart[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1780 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1781 print LSU substr(@seqDNA[$best], @lsuStart[$best] - 1, @lsuEnd[$best] - @lsuStart[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1782 #if (@lsuEnd[$best] - @lsuStart[$best] + 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1783 # print LSU substr(@seqDNA[$best], @lsuEnd[$best] - @lsuStart[$best] + 1, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1784 #} else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1785 # print LSU substr(@seqDNA[$best], @lsuEnd[$best] - @lsuStart[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1786 #}
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1787 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1788 print LSU substr(@seqDNA[$best], @lsuStart[$best] - 1, @lsuEnd[$best] - @lsuStart[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1789 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1790 print LSU "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1791 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1792 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1793 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1794
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1795 ## Write 5.8S sequence to file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1796 if ($out_58S == 1) { # If 5.8S output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1797 if (@problemCode[$best] !~ m/[125]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1798 $seqPartLen = @midEnd[$best] - @midStart[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1799 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1800 print MID $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1801 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1802 print MID ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|5.8S " . "Extracted 5.8S sequence (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1803 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1804 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1805 if (@midStart[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1806 print MID substr(@seqDNA[$best], @midStart[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1807 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1808 print MID substr(@seqDNA[$best], 0, @midStart[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1809 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1810 print MID substr(@seqDNA[$best], @midStart[$best] - 1, @midEnd[$best] - @midStart[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1811 if (@midEnd[$best] - @midStart[$best] + 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1812 print MID substr(@seqDNA[$best], @midEnd[$best], $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1813 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1814 print MID substr(@seqDNA[$best], @midEnd[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1815 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1816 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1817 print MID substr(@seqDNA[$best], @midStart[$best] - 1, @midEnd[$best] - @midStart[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1818 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1819 print MID "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1820 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1821 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1822
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1823 ## Write ITS1 sequence to file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1824 if ($out_its1 == 1) { # If ITS1 output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1825 $seqPartLen = @its1End[$best] - @its1Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1826 if ($seqPartLen > 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1827 if ( (($only_full == 0) && (@problemCode[$best] !~ m/[15]/)) || (($only_full == 1) && (@problemCode[$best] !~ m/[S15]/))) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1828 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1829 print ITS1 $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1830 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1831 print ITS1 ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS1 " . "Extracted ITS1 sequence " . @its1Start[$best] . "-" . @its1End[$best] . " (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1832 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1833 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1834 if (@its1Start[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1835 print ITS1 substr(@seqDNA[$best], @its1Start[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1836 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1837 print ITS1 substr(@seqDNA[$best], 0, @its1Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1838 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1839 print ITS1 substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1840 if (length(@seqDNA[$best]) - @its1End[$best] - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1841 print ITS1 substr(@seqDNA[$best], @its1End[$best], $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1842 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1843 print ITS1 substr(@seqDNA[$best], @its1End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1844 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1845 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1846 if (@its1Start[$best] - 1 - @allanchorLens[0] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1847 print ITS1 substr(@seqDNA[$best], @its1Start[$best] - 1 - @allanchorLens[0], @allanchorLens[0]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1848 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1849 print ITS1 substr(@seqDNA[$best], 0, @its1Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1850 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1851 print ITS1 substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1852 if (length(@seqDNA[$best]) - @its1End[$best] - @allanchorLens[1] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1853 print ITS1 substr(@seqDNA[$best], @its1End[$best], @allanchorLens[1]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1854 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1855 print ITS1 substr(@seqDNA[$best], @its1End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1856 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1857
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1858 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1859 print ITS1 "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1860 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1861 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1862 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1863 if (@problemCode[$best] !~ m/[S15]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1864 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1865 print ITS1PARTIAL $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1866 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1867 print ITS1PARTIAL ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS1 " . "Extracted Full ITS1 sequence " . @its1Start[$best] . "-" . @its1End[$best] . " (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1868 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1869 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1870 if (@its1Start[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1871 print ITS1PARTIAL substr(@seqDNA[$best], @its1Start[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1872 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1873 print ITS1PARTIAL substr(@seqDNA[$best], 0, @its1Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1874 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1875 print ITS1PARTIAL substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1876 if (length(@seqDNA[$best]) - @its1End[$best] - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1877 print ITS1PARTIAL substr(@seqDNA[$best], @its1End[$best], $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1878 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1879 print ITS1PARTIAL substr(@seqDNA[$best], @its1End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1880 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1881 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1882 if (@its1Start[$best] - 1 - @allanchorLens[0] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1883 print ITS1PARTIAL substr(@seqDNA[$best], @its1Start[$best] - 1 - @allanchorLens[0], @allanchorLens[0]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1884 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1885 print ITS1PARTIAL substr(@seqDNA[$best], 0, @its1Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1886 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1887 print ITS1PARTIAL substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1888 if (length(@seqDNA[$best]) - @its1End[$best] - @allanchorLens[1] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1889 print ITS1PARTIAL substr(@seqDNA[$best], @its1End[$best], @allanchorLens[1]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1890 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1891 print ITS1PARTIAL substr(@seqDNA[$best], @its1End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1892 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1893
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1894 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1895 print ITS1PARTIAL "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1896 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1897 if ($out_partial < $seqPartLen) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1898 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1899 print ITS1PARTIAL $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1900 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1901 print ITS1PARTIAL ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS1 " . "Extracted Partial ITS1 sequence " . @its1Start[$best] . "-" . @its1End[$best] . " (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1902 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1903 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1904 if (@its1Start[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1905 print ITS1PARTIAL substr(@seqDNA[$best], @its1Start[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1906 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1907 print ITS1PARTIAL substr(@seqDNA[$best], 0, @its1Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1908 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1909 print ITS1PARTIAL substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1910 if (length(@seqDNA[$best]) - @its1End[$best] - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1911 print ITS1PARTIAL substr(@seqDNA[$best], @its1End[$best], $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1912 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1913 print ITS1PARTIAL substr(@seqDNA[$best], @its1End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1914 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1915 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1916 if (@its1Start[$best] - 1 - @allanchorLens[0] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1917 print ITS1PARTIAL substr(@seqDNA[$best], @its1Start[$best] - 1 - @allanchorLens[0], @allanchorLens[0]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1918 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1919 print ITS1PARTIAL substr(@seqDNA[$best], 0, @its1Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1920 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1921 print ITS1PARTIAL substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1922 if (length(@seqDNA[$best]) - @its1End[$best] - @allanchorLens[1] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1923 print ITS1PARTIAL substr(@seqDNA[$best], @its1End[$best], @allanchorLens[1]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1924 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1925 print ITS1PARTIAL substr(@seqDNA[$best], @its1End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1926 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1927 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1928 print ITS1PARTIAL "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1929 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1930 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1931 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1932 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1933
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1934 ## Write ITS2 sequence to file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1935 if ($out_its2 == 1) { # If ITS2 output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1936 $seqPartLen = @its2End[$best] - @its2Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1937 if ($seqPartLen > 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1938 if ( (($only_full == 0) && (@problemCode[$best] !~ m/[25]/)) || (($only_full == 1) && (@problemCode[$best] !~ m/[L25]/))) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1939 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1940 print ITS2 $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1941 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1942 print ITS2 ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS2 " . "Extracted ITS2 sequence " . @its2Start[$best] . "-" . @its2End[$best] . " (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1943 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1944 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1945 if (@its2Start[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1946 print ITS2 substr(@seqDNA[$best], @its2Start[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1947 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1948 print ITS2 substr(@seqDNA[$best], 0, @its2Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1949 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1950 print ITS2 substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1951 if (length(@seqDNA[$best]) - @its2End[$best] - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1952 print ITS2 substr(@seqDNA[$best], @its2End[$best], $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1953 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1954 print ITS2 substr(@seqDNA[$best], @its2End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1955 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1956 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1957 if (@its2Start[$best] - 1 - @allanchorLens[2] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1958 print ITS2 substr(@seqDNA[$best], @its2Start[$best] - 1 - @allanchorLens[2], @allanchorLens[2]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1959 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1960 print ITS2 substr(@seqDNA[$best], 0, @its2Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1961 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1962 print ITS2 substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1963 if (length(@seqDNA[$best]) - @its2End[$best] - @allanchorLens[3] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1964 print ITS2 substr(@seqDNA[$best], @its2End[$best], @allanchorLens[3]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1965 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1966 print ITS2 substr(@seqDNA[$best], @its2End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1967 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1968 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1969 print ITS2 "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1970 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1971 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1972 if (@problemCode[$best] !~ m/[L25]/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1973 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1974 print ITS2PARTIAL $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1975 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1976 print ITS2PARTIAL ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS2 " . "Extracted Full ITS2 sequence " . @its2Start[$best] . "-" . @its2End[$best] . " (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1977 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1978 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1979 if (@its2Start[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1980 print ITS2PARTIAL substr(@seqDNA[$best], @its2Start[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1981 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1982 print ITS2PARTIAL substr(@seqDNA[$best], 0, @its2Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1983 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1984 print ITS2PARTIAL substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1985 if (length(@seqDNA[$best]) - @its2End[$best] - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1986 print ITS2PARTIAL substr(@seqDNA[$best], @its2End[$best], $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1987 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1988 print ITS2PARTIAL substr(@seqDNA[$best], @its2End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1989 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1990 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1991 if (@its2Start[$best] - 1 - @allanchorLens[2] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1992 print ITS2PARTIAL substr(@seqDNA[$best], @its2Start[$best] - 1 - @allanchorLens[2], @allanchorLens[2]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1993 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1994 print ITS2PARTIAL substr(@seqDNA[$best], 0, @its2Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1995 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1996 print ITS2PARTIAL substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1997 if (length(@seqDNA[$best]) - @its2End[$best] - @allanchorLens[3] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1998 print ITS2PARTIAL substr(@seqDNA[$best], @its2End[$best], @allanchorLens[3]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
1999 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2000 print ITS2PARTIAL substr(@seqDNA[$best], @its2End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2001 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2002 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2003 print ITS2PARTIAL "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2004 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2005 if ($out_partial < $seqPartLen) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2006 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2007 print ITS2PARTIAL $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2008 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2009 print ITS2PARTIAL ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS2 " . "Extracted Partial ITS2 sequence " . @its2Start[$best] . "-" . @its2End[$best] . " (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2010 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2011 if (uc($anchor) ne "HMM") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2012 if (@its2Start[$best] - 1 - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2013 print ITS2PARTIAL substr(@seqDNA[$best], @its2Start[$best] - 1 - $anchorLen, $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2014 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2015 print ITS2PARTIAL substr(@seqDNA[$best], 0, @its2Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2016 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2017 print ITS2PARTIAL substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2018 if (length(@seqDNA[$best]) - @its2End[$best] - $anchorLen > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2019 print ITS2PARTIAL substr(@seqDNA[$best], @its2End[$best], $anchorLen);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2020 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2021 print ITS2PARTIAL substr(@seqDNA[$best], @its2End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2022 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2023 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2024 if (@its2Start[$best] - 1 - @allanchorLens[2] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2025 print ITS2PARTIAL substr(@seqDNA[$best], @its2Start[$best] - 1 - @allanchorLens[2], @allanchorLens[2]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2026 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2027 print ITS2PARTIAL substr(@seqDNA[$best], 0, @its2Start[$best] - 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2028 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2029 print ITS2PARTIAL substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2030 if (length(@seqDNA[$best]) - @its2End[$best] - @allanchorLens[3] > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2031 print ITS2PARTIAL substr(@seqDNA[$best], @its2End[$best], @allanchorLens[3]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2032 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2033 print ITS2PARTIAL substr(@seqDNA[$best], @its2End[$best]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2034 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2035 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2036 print ITS2PARTIAL "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2037 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2038 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2039 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2040 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2041 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2042
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2043 ## Output concatenated ITS1 + ITS2 sequences
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2044 if ($out_concat == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2045 $seqPartLen1 = @its1End[$best] - @its1Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2046 $seqPartLen2 = @its2End[$best] - @its2Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2047 if (($seqPartLen1 >= $concat_minlen) && ($seqPartLen2 >= $concat_minlen)) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2048 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2049 print CONCAT $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2050 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2051 print CONCAT ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS1+2 " . "Concatenated ITS1 and ITS2 sequences (" . ($seqPartLen1 + $seqPartLen2) . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2052 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2053 print CONCAT substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1); # Write ITS1 DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2054 print CONCAT "-----"; # Write spacer
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2055 print CONCAT substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1); # Write ITS2 DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2056 print CONCAT "\n"; # Write newline
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2057 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2058 if ($seqPartLen1 >= $concat_minlen) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2059 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2060 print CONCAT $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2061 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2062 print CONCAT ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS1 " . "ITS1 sequence (ITS2 too short) (" . $seqPartLen1 . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2063 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2064 print CONCAT substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1); # Write ITS1 DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2065 print CONCAT "-----"; # Write spacer
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2066 print CONCAT "\n"; # Write newline
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2067 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2068 if ($seqPartLen2 >= $concat_minlen) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2069 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2070 print CONCAT $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2071 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2072 print CONCAT ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS2 " . "ITS2 sequence (ITS1 too short) (" . $seqPartLen2 . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2073 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2074 print CONCAT "-----"; # Write spacer
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2075 print CONCAT substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1); # Write ITS2 DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2076 print CONCAT "\n"; # Write newline
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2077 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2078 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2079 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2080
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2081 ## Print all sequences to the joined file for debugging
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2082 if ($out_joined == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2083 $seqPartLen = @ssuEnd[$best] - @ssuStart[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2084 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2085 print JOINED $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2086 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2087 print JOINED ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|SSU " . "Extracted SSU sequence (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2088 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2089 print JOINED substr(@seqDNA[$best], @ssuStart[$best] - 1, @ssuEnd[$best] - @ssuStart[$best] + 1) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2090
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2091 $seqPartLen = @its1End[$best] - @its1Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2092 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2093 print JOINED $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2094 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2095 print JOINED ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS1 " . "Extracted ITS1 sequence " . @its1Start[$best] . "-" . @its1End[$best] . " (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2096 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2097 print JOINED substr(@seqDNA[$best], @its1Start[$best] - 1, @its1End[$best] - @its1Start[$best] + 1) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2098
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2099 $seqPartLen = @midEnd[$best] - @midStart[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2100 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2101 print JOINED $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2102 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2103 print JOINED ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|5.8S " . "Extracted 5.8S sequence (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2104 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2105 print JOINED substr(@seqDNA[$best], @midStart[$best] - 1, @midEnd[$best] - @midStart[$best] + 1) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2106
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2107 $seqPartLen = @its2End[$best] - @its2Start[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2108 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2109 print JOINED $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2110 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2111 print JOINED ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|ITS2 " . "Extracted ITS2 sequence (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2112 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2113 print JOINED substr(@seqDNA[$best], @its2Start[$best] - 1, @its2End[$best] - @its2Start[$best] + 1) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2114
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2115 $seqPartLen = @lsuEnd[$best] - @lsuStart[$best] + 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2116 if ($out_preserve == 1) { # If sequence headers should be preserved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2117 print JOINED $headers{@seqID[$best]} . "\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2118 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2119 print JOINED ">" . @seqID[$best] . "|" . @seqITSType[$best] . "|LSU " . "Extracted LSU sequence (" . $seqPartLen . " bp)\n"; # Write FASTA definition line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2120 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2121 print JOINED substr(@seqDNA[$best], @lsuStart[$best] - 1, @lsuEnd[$best] - @lsuStart[$best] + 1) . "\n"; # Write DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2122 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2123
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2124
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2125 if ($out_results == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2126 ## Print info on all matches, also not top ones to the results file...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2127 for ($i = 0; $i < scalar(@seqITSType); $i++) { # Go through all the possible ITS types...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2128 print RESULTS @seqITSType[$i] . ": " . @seqDomCounts[$i] . " " . @seqAvgE[$i] . " " . @seqAvgScore[$i]; # Write some info on this type (Type, Domain count, Average E-value, Average score)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2129 if ($i < scalar(@seqITSType) - 1) { # If this is no the last domain type
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2130 print RESULTS ", "; # Write a comma
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2131 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2132 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2133 print RESULTS "\n"; # Write end of line
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2134 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2135 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2136
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2137 ## Undefine all used arrays for the next round...
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2138 undef @seqID;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2139 undef @seqITSType;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2140 undef @seqCompl;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2141 undef @seqDomCounts;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2142 undef @seqAvgE;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2143 undef @seqAvgScore;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2144 undef @seqScoreSum;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2145 undef @seqDNA;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2146 undef @allSeqDomains;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2147 undef @dnaStart;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2148 undef @dnaEnd;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2149 undef @ssuStart;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2150 undef @ssuEnd;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2151 undef @lsuStart;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2152 undef @lsuEnd;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2153 undef @midStart;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2154 undef @midEnd;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2155 undef @its1Start;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2156 undef @its1End;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2157 undef @its2Start;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2158 undef @its2End;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2159 undef @startDomain;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2160 undef @endDomain;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2161 undef @domain_order;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2162 undef @sorted_domain_order;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2163 undef @problem;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2164 undef @problemCode;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2165 undef @anchorLens;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2166 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2167 if (@item[0] ne "") { # Add this entry to the set (regardless if the entry has the same ID as entries already in the set), as long as it is non-empty
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2168 push(@seqID, @item[0]); # Add sequence ID
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2169 push(@seqITSType, @item[1]); # Add ITS type
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2170 push(@seqCompl, @item[2]); # Add main/complementary strand info
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2171 push(@seqDomCounts, @item[3]); # Add domain count
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2172 push(@seqAvgE, @item[4]); # Add average E-value
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2173 push(@seqAvgScore, @item[5]); # Add average score
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2174 push(@seqScoreSum, @item[6]); # Add sum-of-scores
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2175 push(@seqDNA, @item[7]); # Add DNA sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2176
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2177 ## Determine first and last domains, and their positions
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2178 ## Set variables to unrealistic values
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2179 $dnaEnd = length(@item[7]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2180 $startDomain = "***";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2181 $endDomain = "***";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2182 $dnaStart = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2183 $allDomains = "";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2184 $ssuStart = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2185 $ssuEnd = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2186 $midStart = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2187 $midEnd = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2188 $lsuStart = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2189 $lsuEnd = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2190 $its1Start = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2191 $its1End = length(@item[7]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2192 $its2Start = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2193 $its2End = length(@item[7]);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2194 $problem = "";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2195 $problemCode = "";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2196
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2197 $ssuFound = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2198 $lsuFound = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2199 $midSFound = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2200 $midEFound = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2201 $order = "";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2202 undef @hitanchorlens;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2203
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2204 for ($i = 8; $i < scalar(@item); $i++) { # Go through the list of found domains in this sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2205 ($hitStart,$hitEnd,$hitProfile,$hitScore,$hitEval,$hitanchorlen) = split(';',@item[$i]); # Separate the hit stats into variables
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2206 $allDomains = $allDomains . $hitProfile . " "; # Add found domain to the list of all domains
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2207
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2208 # if ($hitStart < $dnaStart) { # If this domain is the first one so far
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2209 # $dnaStart = $hitStart; # Set the start of the ITS sequence to this domain's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2210 # $startDomain = $hitProfile; # Set this domain as the starting domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2211 # }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2212 # if ($hitEnd > $dnaEnd) { # If this domain is the last one so far
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2213 # $dnaEnd = $hitEnd; # Set the end of the ITS sequence to this domain's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2214 # $endDomain = $hitProfile; # Set this domain as the ending domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2215 # }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2216
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2217 if (substr($hitProfile,0,5) eq "1_SSU" ) { # If this domain is the SSU's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2218 $dnaStart = $hitStart; # Set the start of the ITS sequence to this domain's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2219 $startDomain = $hitProfile; # Set this domain as the starting domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2220 $ssuStart = 1; # Set the start of the SSU sequence to the start of the sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2221 $ssuEnd = $hitEnd; # Set the end of the SSU sequence to this domain's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2222 $its1Start = $hitEnd + 1; # Set the start of the ITS1 sequence to right after this domain's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2223 if ($midEFound == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2224 $its2Start = $hitEnd + 1; # Set the end of the ITS2 sequence to right after this domain's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2225 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2226 $ssuFound = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2227 $order = $order . "1";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2228 @hitanchorlens[1] = $hitanchorlen;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2229 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2230
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2231 if (substr($hitProfile,0,5) eq "2_5.8" ) { # If this domain is the 5.8S's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2232 if ($startDomain eq "***") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2233 $dnaStart = 1; # Set the start of the ITS sequence to the start of the sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2234 $startDomain = $hitProfile; # Set this domain as the starting domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2235 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2236 $its1End = $hitStart - 1; # Set the end of the ITS1 sequence to right before this domain's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2237 if ($midEFound == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2238 $its2Start = $hitEnd + 1; # Set the start of the ITS2 sequence to right after this domain's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2239 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2240 $midStart = $hitStart; # Set the start of the 5.8S sequence to this domain's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2241 $midSFound = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2242 $order = $order . "2";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2243 @hitanchorlens[2] = $hitanchorlen;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2244 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2245
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2246 if (substr($hitProfile,0,5) eq "3_End" ) { # If this domain is the 5.8S's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2247 if ($startDomain eq "***") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2248 $dnaStart = 1; # Set the start of the ITS sequence to the start of the sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2249 $startDomain = $hitProfile; # Set this domain as the starting domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2250 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2251 if ($midSFound == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2252 $its1End = $hitStart - 1; # Set the end of the ITS1 sequence to right before this domain's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2253 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2254 $its2Start = $hitEnd + 1; # Set the end of the ITS2 sequence to right after this domain's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2255 $midEnd = $hitEnd; # Set the end of the 5.8S sequence to this domain's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2256 $midEFound = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2257 $order = $order . "3";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2258 @hitanchorlens[3] = $hitanchorlen;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2259 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2260
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2261 if (substr($hitProfile,0,5) eq "4_LSU" ) { # If this domain is the LSU's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2262 if ($startDomain eq "***") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2263 $dnaStart = 1; # Set the start of the ITS sequence to the start of the sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2264 $startDomain = $hitProfile; # Set this domain as the starting domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2265 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2266 $dnaEnd = $hitEnd; # Set the end of the ITS sequence to this domain's end
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2267 $endDomain = $hitProfile; # Set this domain as the ending domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2268 $lsuStart = $hitStart; # Set the start of the LSU sequence to the start of the this domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2269 $lsuEnd = length(@item[7]); # Set the end of the LSU sequence to the end of the sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2270 $its2End = $hitStart - 1; # Set the end of the ITS2 sequence to right before this domain's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2271 if ($midSFound == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2272 $its1End = $hitStart - 1; # Set the end of the ITS1 sequence to right before this domain's start
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2273 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2274 $lsuFound = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2275 $order = $order . "4";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2276 @hitanchorlens[4] = $hitanchorlen;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2277 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2278 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2279
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2280 if ($ssuFound == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2281 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2282 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2283 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2284 $problem = $problem . "End of SSU sequence not found";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2285 $problemCode = $problemCode . "S";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2286 $order = "1" . $order;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2287 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2288
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2289 if ($lsuFound == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2290 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2291 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2292 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2293 $problem = $problem . "Start of LSU sequence not found";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2294 $problemCode = $problemCode . "L";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2295 $order = $order . "4";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2296 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2297
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2298 if (($midSFound == 0) && ($midEFound == 0)) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2299 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2300 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2301 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2302 $problem = $problem . "The 5.8S sequence was not found at all";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2303 $problemCode = $problemCode . "5";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2304 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2305 if ($midSFound == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2306 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2307 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2308 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2309 $problem = $problem . "Start of 5.8S sequence not found";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2310 $problemCode = $problemCode . "1";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2311 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2312 if ($midEFound == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2313 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2314 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2315 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2316 $problem = $problem . "End of 5.8S sequence not found";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2317 $problemCode = $problemCode . "2";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2318 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2319 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2320
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2321 if ($dnaEnd - $dnaStart > 1500) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2322 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2323 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2324 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2325 $problem = $problem . "ITS region is suspiciously long (> 1500 bp)";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2326 $problemCode = $problemCode . "B";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2327 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2328
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2329 if (length($order) == 4) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2330 if ($order ne "1234") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2331 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2332 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2333 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2334 $problem = $problem . "Domains found in wrong order, sequence may be chimeric";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2335 $problemCode = $problemCode . "C";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2336 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2337 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2338 if ((substr($order,0,1) ne "1") || (substr($order,-1) ne "4")) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2339 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2340 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2341 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2342 $problem = $problem . "Domains found in wrong order, sequence may be chimeric";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2343 $problemCode = $problemCode . "C";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2344 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2345 if (($midSFound == 1) && ($midEFound == 1)) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2346 if ($order !~ m/23/) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2347 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2348 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2349 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2350 $problem = $problem . "Domains found in wrong order, sequence may be chimeric";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2351 $problemCode = $problemCode . "C";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2352 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2353 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2354 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2355 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2356
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2357 if ($its1End - $its1Start < 0) { # 5.8S overlaps SSU
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2358 if (($midSFound == 1) && ($ssuFound == 1)) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2359 $problemCode = $problemCode . "OC";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2360 $its1Start = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2361 $its1End = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2362 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2363 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2364 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2365 $problem = $problem . "SSU seems to overlap 5.8S, sequence may be chimeric";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2366 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2367 $problemCode = $problemCode . "X";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2368 $its1Start = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2369 $its1End = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2370 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2371 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2372 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2373 $problem = $problem . "No ITS1 sequence";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2374 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2375 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2376 if ($its2End - $its2Start < 0) { # 5.8S overlaps LSU
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2377 if (($midEFound == 1) && ($lsuFound == 1)) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2378 $problemCode = $problemCode . "PC";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2379 $its2Start = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2380 $its2End = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2381 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2382 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2383 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2384 $problem = $problem . "LSU seems to overlap 5.8S, sequence may be chimeric";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2385 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2386 $problemCode = $problemCode . "Y";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2387 $its2Start = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2388 $its2End = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2389 if ($problem ne "") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2390 $problem = $problem . "; ";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2391 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2392 $problem = $problem . "No ITS2 sequence";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2393 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2394 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2395
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2396
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2397 if (($problemCode =~ m/[15]/) && ($problemCode =~ m/[S]/)) { # Sequence lack all indications of ITS1
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2398 $its1Start = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2399 $its1End = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2400 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2401 if (($problemCode =~ m/[25]/) && ($problemCode =~ m/[L]/)) { # Sequence lack all indications of ITS2
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2402 $its2Start = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2403 $its2End = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2404 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2405
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2406
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2407 $allhitanchorlens = @hitanchorlens[1] . "," . @hitanchorlens[2] . "," . @hitanchorlens[3] . "," . @hitanchorlens[4];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2408
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2409 push(@allSeqDomains,$allDomains); # Add list of all domains
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2410 push(@dnaStart, $dnaStart); # Add start of the ITS sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2411 push(@startDomain, $startDomain); # Add start domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2412 push(@dnaEnd, $dnaEnd); # Add end of ITS sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2413 push(@endDomain, $endDomain); # Add end domain
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2414 push(@ssuStart, $ssuStart); # Add start of the SSU sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2415 push(@ssuEnd, $ssuEnd); # Add end of the SSU sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2416 push(@midStart, $midStart); # Add start of the 5.8S sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2417 push(@midEnd, $midEnd); # Add end of the 5.8S sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2418 push(@lsuStart, $lsuStart); # Add start of the LSU sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2419 push(@lsuEnd, $lsuEnd); # Add end of the LSU sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2420 push(@its1Start, $its1Start); # Add start of the ITS1 sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2421 push(@its1End, $its1End); # Add end of the ITS1 sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2422 push(@its2Start, $its2Start); # Add start of the ITS2 sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2423 push(@its2End, $its2End); # Add end of the ITS2 sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2424 push(@problem, $problem); # Add potential problem info
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2425 push(@problemCode, $problemCode); # Add potential problem info in code form
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2426 push(@anchorLens, $allhitanchorlens); # Add the anchor lengths
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2427 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2428 $lc++; # Increase the line count by one
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2429 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2430
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2431 ## Save results to the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2432 if ($out_sum == 1) { # If summary file should be written
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2433 $itsTotal = 0; # Reset the total ITS sum
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2434 foreach $typeCount (@itsCounts) { # Add ITSs from all different origins
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2435 $itsTotal += $typeCount; # Add the number of ITSs for this origin
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2436 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2437 ## Write info on the found ITS sequences to the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2438 print SUMMARY "Sequences detected as ITS by ITSx:\t$itsTotal\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2439 print SUMMARY " On main strand: \t$itsMain\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2440 print SUMMARY " On complementary strand:\t$itsCompl\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2441 if ($allow_reorder == 0) { # If re-ordering of domains is not allowed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2442 print SUMMARY "Sequences detected as chimeric by ITSx:\t$itsChimeric\n"; # Write the number of reported chimeric sequences to the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2443 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2444 ## Write info on the found ITS sequence types to the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2445 print SUMMARY "ITS sequences by preliminary origin:\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2446 print SUMMARY " Alveolates: \t" . int(@itsCounts[ord("A")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2447 print SUMMARY " Amoebozoa: \t" . int(@itsCounts[ord("D")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2448 print SUMMARY " Bacillariophyta: \t" . int(@itsCounts[ord("C")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2449 print SUMMARY " Brown algae: \t" . int(@itsCounts[ord("I")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2450 print SUMMARY " Bryophytes: \t" . int(@itsCounts[ord("B")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2451 print SUMMARY " Euglenozoa: \t" . int(@itsCounts[ord("E")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2452 print SUMMARY " Eustigmatophytes:\t" . int(@itsCounts[ord("U")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2453 print SUMMARY " Fungi: \t" . int(@itsCounts[ord("F")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2454 print SUMMARY " Green algae: \t" . int(@itsCounts[ord("G")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2455 print SUMMARY " Liverworts: \t" . int(@itsCounts[ord("L")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2456 print SUMMARY " Metazoa: \t" . int(@itsCounts[ord("M")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2457 print SUMMARY " Microsporidia: \t" . int(@itsCounts[ord("N")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2458 print SUMMARY " Oomycetes: \t" . int(@itsCounts[ord("O")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2459 print SUMMARY " Prymnesiophytes: \t" . int(@itsCounts[ord("P")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2460 print SUMMARY " Raphidophytes: \t" . int(@itsCounts[ord("Q")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2461 print SUMMARY " Red algae: \t" . int(@itsCounts[ord("H")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2462 print SUMMARY " Rhizaria: \t" . int(@itsCounts[ord("R")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2463 print SUMMARY " Synurophyceae: \t" . int(@itsCounts[ord("S")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2464 print SUMMARY " Tracheophyta: \t" . int(@itsCounts[ord("T")]) . "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2465 print SUMMARY "-----------------------------------------------------------------\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2466 close (SUMMARY); # Close the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2467 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2468
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2469 if ($out_results == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2470 close (RESULTS); # Close the results file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2471 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2472 close (RAWOUT); # Close the raw output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2473 close (PROBLEM); # Close the file for problematic entries
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2474 if ($foundProblem == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2475 `rm $output.problematic.txt 2> /dev/null`;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2476 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2477 if ($out_pos == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2478 close (POS); # Close the positions file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2479 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2480 if ($out_fasta == 1) { # If FASTA output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2481 close (FASTA); # Close the FASTA output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2482 if ($allow_reorder == 0) { # If re-ordering of domains is not allowed
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2483 close (CHIMERA); # Close the chimera file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2484 if ($foundChimera == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2485 `rm $output.chimeric.fasta 2> /dev/null`;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2486 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2487 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2488 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2489 close FULLPARTIAL;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2490 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2491 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2492 if ($out_ssu == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2493 close (SSU); # Close the SSU file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2494 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2495 if ($out_lsu == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2496 close (LSU); # Close the LSU file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2497 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2498 if ($out_58S == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2499 close (MID); # Close the 58S file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2500 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2501 if ($out_its1 == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2502 close (ITS1); # Close the ITS1 file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2503 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2504 close ITS1PARTIAL;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2505 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2506 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2507 if ($out_its2 == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2508 close (ITS2); # Close the ITS2 file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2509 if ($out_partial > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2510 close ITS2PARTIAL;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2511 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2512 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2513 if ($out_concat == 1) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2514 close CONCAT;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2515 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2516
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2517 if ($out_not == 1) { # If not-found output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2518 open (NOTFOUND, "$output\_no_detections.txt"); # Open the not-found output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2519 while ($line = <NOTFOUND>) { # Read all entries from file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2520 chomp($line); # Remove newline char
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2521 push(@nodetectionlist,$line); # Add to non-detection list
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2522 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2523 close NOTFOUND;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2524
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2525 open (NOTFOUND, ">$output\_no_detections.fasta"); # Create a not-found FASTA output file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2526 foreach $seqID (@nodetectionlist) { # For all non-detections
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2527 $seq = $sequenceDB{"$seqID"}; # Get sequence from sequence database
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2528 print NOTFOUND ">$seqID\n"; # Print not found sequence ID
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2529 print NOTFOUND $seq . "\n"; # Print not found sequence
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2530 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2531 close NOTFOUND;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2532 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2533
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2534 ## Clean up and finish
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2535
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2536 if ($pipeline == 0) { # If ITSx is not called from the pipeline mode (i.e. from ITSx)
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2537 if ($save_raw == 1) { # If raw data should be saved
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2538 `mv $tempDir $output\_ITSx_raw_output`; # Change the name of the temporary directory to ..._ITSx_raw_output
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2539 } else { # Else, discard the raw data
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2540 `rm -rf $tempDir`; # Remove the temporary directory
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2541 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2542 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2543
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2544 ## Get the current time and output a finished message
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2545 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2546 if ($silent == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2547 print STDERR "$now : Extraction finished!\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2548 print STDERR "-----------------------------------------------------------------\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2549 print STDERR "Thank you for using ITSx!\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2550 print STDERR "Please report bugs or unsupported lineages to itsx\@microbiology.se\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2551 print STDERR "\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2552 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2553
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2554 ## Write end time a summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2555 if ($pipeline == 0) { # If not running in pipeline mode
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2556 if ($out_sum == 1) { # If summary output is on
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2557 open (SUMMARY, ">>$output.summary.txt"); # Append to the summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2558 print SUMMARY "ITSx run finished at $now.\n"; # Write ending time for the analysis
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2559 close (SUMMARY); # Close summary file
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2560 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2561 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2562
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2563 sub hmmerSearch {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2564 $hmmerCommand = $_[0];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2565 $outputFile = $_[1];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2566 $strand = $_[2];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2567 $profileSet = $_[3];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2568 open (HMMEROUTPUT, ">$outputFile");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2569 open (HMMER, "$hmmerCommand |");
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2570 $totalHitCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2571 $hitCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2572 $SSUCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2573 $LSUCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2574 $startCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2575 $endCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2576 $maxCount = 1;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2577 while (chomp($line = <HMMER>)) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2578 if (substr($line,0,6) eq "Query:") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2579 $hitCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2580 $SSUCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2581 $LSUCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2582 $startCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2583 $endCount = 0;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2584 undef @bestScore;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2585 undef @bestEntry;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2586
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2587 $query = substr($line,7);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2588 $queryLength = $query;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2589 $queryLength =~ s/.* *//;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2590 $queryLength =~ s/[^0-9]//g;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2591 $query =~ s/ *//;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2592 $query =~ s/ *.*//;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2593
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2594 print HMMEROUTPUT "## New query:\t$query\t$queryLength\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2595
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2596 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2597 if (substr($line,0,12) eq "Description:") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2598 $desc = $line;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2599 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2600 if (substr($line,0,3) eq ">> ") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2601 ($tempshit,$hmmerSubjectName) = split(' ',$line);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2602 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2603 if ($line =~ m/[0-9] ! /) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2604 $stats = $line;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2605 $stats =~ s/ */\t/g;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2606 ($empty,$no,$excl,$score) = split('\t',$stats);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2607 $hitCount++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2608 $totalHitCount++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2609 if ($maxCount == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2610 print HMMEROUTPUT "$query\t$hmmerSubjectName\t$queryLength$stats\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2611 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2612 if (substr($hmmerSubjectName,0,5) eq "1_SSU") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2613 $SSUCount++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2614 if ($SSUCount <= $maxCount) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2615 @bestScore[1] = $score;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2616 @bestEntry[1] = "$query\t$hmmerSubjectName\t$queryLength$stats\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2617 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2618 if ($score > @bestScore[1]) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2619 @bestScore[1] = $score;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2620 @bestEntry[1] = "$query\t$hmmerSubjectName\t$queryLength$stats\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2621 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2622 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2623 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2624 if (substr($hmmerSubjectName,0,5) eq "4_LSU") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2625 $LSUCount++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2626 if ($LSUCount <= $maxCount) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2627 @bestScore[4] = $score;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2628 @bestEntry[4] = "$query\t$hmmerSubjectName\t$queryLength$stats\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2629 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2630 if ($score > @bestScore[4]) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2631 @bestScore[4] = $score;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2632 @bestEntry[4] = "$query\t$hmmerSubjectName\t$queryLength$stats\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2633 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2634 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2635 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2636 if (substr($hmmerSubjectName,0,5) eq "2_5.8") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2637 $startCount++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2638 if ($startCount <= $maxCount) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2639 @bestScore[2] = $score;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2640 @bestEntry[2] = "$query\t$hmmerSubjectName\t$queryLength$stats\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2641 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2642 if ($score > @bestScore[2]) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2643 @bestScore[2] = $score;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2644 @bestEntry[2] = "$query\t$hmmerSubjectName\t$queryLength$stats\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2645 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2646 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2647 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2648 if (substr($hmmerSubjectName,0,5) eq "3_End") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2649 $endCount++;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2650 if ($endCount <= $maxCount) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2651 @bestScore[3] = $score;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2652 @bestEntry[3] = "$query\t$hmmerSubjectName\t$queryLength$stats\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2653 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2654 if ($score > @bestScore[3]) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2655 @bestScore[3] = $score;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2656 @bestEntry[3] = "$query\t$hmmerSubjectName\t$queryLength$stats\n"
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2657 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2658 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2659 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2660 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2661 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2662 if (substr($line,0,2) eq "//") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2663 if ($maxCount > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2664 if ($SSUCount > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2665 print HMMEROUTPUT @bestEntry[1];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2666 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2667 if ($startCount > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2668 print HMMEROUTPUT @bestEntry[2];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2669 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2670 if ($endCount > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2671 print HMMEROUTPUT @bestEntry[3];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2672 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2673 if ($LSUCount > 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2674 print HMMEROUTPUT @bestEntry[4];
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2675 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2676 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2677 print HMMEROUTPUT "//\n";
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2678 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2679 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2680 close(HMMEROUTPUT);
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2681 $now = localtime;
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2682 if ($silent == 0) {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2683 if ($strand eq "M") {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2684 print STDERR " $now : " . ucfirst($profileIndex{$profileSet}) . " analysis of main strand finished.\n"; # Print finished type
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2685 } else {
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2686 print STDERR " $now : " . ucfirst($profileIndex{$profileSet}) . " analysis of complementary strand finished.\n"; # Print finished type
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2687 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2688 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2689 }
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2690
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2691
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2692 ## Please send beers, pizzas, cakes, fruit pies, job positions and other types of feedback to:
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2693 ## johan.bengtsson [at] microbiology.se
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2694 ## Looking forward to hearing from you.... visit my website: www.microbiology.se for info on my research
f82c70f54bd7 Uploaded
okorol
parents:
diff changeset
2695 ## //Johan Bengtsson, 2012-2014