annotate rapsodyn/ParseBlastForUniqueMatch.pl @ 7:3f7b0788a1c4 draft

Uploaded
author mcharles
date Tue, 07 Oct 2014 10:34:34 -0400
parents 442a7c88b886
children 0a6c1cfe4dc8
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
1 #!/usr/bin/perl
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
2 #V1.0.1 added log, option parameters
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
3 use strict;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
4 use warnings;
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
5 use Getopt::Long;
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
6
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
7
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
8 my $input_variant_file;
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
9 my $input_blast_file;
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
10 my $log_file;
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
11 my $WINDOW_LENGTH= 50;
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
12 my $NB_MISMATCH_MAX = 3;
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
13
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
14 GetOptions (
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
15 "input_variant_file=s" => \$input_variant_file,
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
16 "input_blast_file=s" => \$input_blast_file,
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
17 "window_length=s" => \$WINDOW_LENGTH,
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
18 "log_file=s" => \$log_file,
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
19 "nb_mismatch_max=s" => \$NB_MISMATCH_MAX
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
20 ) or die("Error in command line arguments\n");
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
21
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
22 my $nb_variant_checked=0;
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
23 my $nb_variant_selected=0;
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
24 my %hash_name;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
25
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
26 open(INB, $input_blast_file) or die ("Can't open $input_blast_file\n");
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
27 while (my $line =<INB>){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
28 my @fields = split (/\s+/,$line);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
29 # print $#fields,"\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
30 # print $line;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
31 # exit(0);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
32 my $query_name;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
33 my $query_start;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
34 my $query_stop;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
35 my $query_aln;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
36 my $subject_aln;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
37 my $compt_mismatch_5p=0;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
38 my $compt_mismatch_3p=0;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
39
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
40 if ($#fields == 24){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
41 $query_name = $fields[0];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
42 $query_start = $fields[6];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
43 $query_stop = $fields[7];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
44 $query_aln = $fields[20];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
45 $subject_aln = $fields[21];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
46 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
47 elsif ($#fields == 4){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
48 $query_name = $fields[0];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
49 $query_start = $fields[1];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
50 $query_stop = $fields[2];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
51 $query_aln = $fields[3];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
52 $subject_aln = $fields[4];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
53 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
54 else {
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
55 print STDERR "Unrecongnized tabular format for blast result\nScript works with 25 column or 5 custom column(qseqid,qstart,qend,ssseq,sseq)\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
56 exit(0);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
57 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
58
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
59
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
60 my @field_name = split (/\_/,$query_name);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
61 my $name = $field_name[0]."_".$field_name[1];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
62 if ($query_name =~ /random/){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
63 $name .= "_".$field_name[2];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
64 # print $name."\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
65
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
66 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
67 if (!$hash_name{$name}){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
68 $hash_name{$name}=0;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
69 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
70 elsif ($hash_name{$name}>1){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
71 next;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
72 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
73 # if ($query_name eq "chrUnn_random_8279117_1_M1_a"){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
74 # print "READ : $query_name\n$name\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
75 # print "HASH : ",$hash_name{$name},"\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
76 # # exit(0);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
77 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
78
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
79
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
80
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
81 chomp($query_aln);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
82 chomp($subject_aln);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
83
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
84 my $nb_gap_query=0;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
85
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
86 if (length($query_aln) == length($subject_aln)){
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
87 if (length($query_aln)<$WINDOW_LENGTH-$NB_MISMATCH_MAX){
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
88 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
89 else {
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
90 my @q = split(//,$query_aln);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
91 my @s = split(//,$subject_aln);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
92 for (my $i=0;$i<=$#q;$i++){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
93 my $global_idx = $query_start-1+$i-$nb_gap_query;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
94 if ($q[$i] eq "-"){
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
95 if ($global_idx < $WINDOW_LENGTH){
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
96 $compt_mismatch_5p++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
97 }
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
98 elsif ($global_idx > $WINDOW_LENGTH){
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
99 $compt_mismatch_3p++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
100 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
101 $nb_gap_query++; #On compte les gap dans la query pour les soustraire de l'index global
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
102 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
103 else {
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
104 if ($q[$i] ne $s[$i]){
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
105 if ($global_idx < $WINDOW_LENGTH){
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
106 $compt_mismatch_5p++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
107 }
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
108 elsif ($global_idx > $WINDOW_LENGTH){
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
109 $compt_mismatch_3p++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
110 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
111 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
112 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
113 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
114 $compt_mismatch_5p += $query_start-1;
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
115 $compt_mismatch_3p += $WINDOW_LENGTH *2 + 1 - $query_stop;
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
116
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
117 # for (my $i=0;$i<$window_length;$i++){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
118 # if ($tbl_q_aln[$i] eq "#"){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
119 # $compt_mismatch_5p++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
120 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
121 # elsif ($tbl_q_aln[$i] ne $tbl_s_aln[$i]){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
122 # $compt_mismatch_5p++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
123 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
124 # else {
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
125 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
126
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
127 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
128 # for (my $i=$window_length+1;$i<=$window_length*2;$i++){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
129 # if ($tbl_q_aln[$i] eq "#"){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
130 # $compt_mismatch_3p++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
131 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
132 # elsif ($tbl_q_aln[$i] ne $tbl_s_aln[$i]){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
133 # $compt_mismatch_3p++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
134 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
135 # else {
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
136 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
137 # }
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
138 if (($compt_mismatch_5p <= $NB_MISMATCH_MAX)||($compt_mismatch_3p <= $NB_MISMATCH_MAX)){
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
139 $hash_name{$name}++;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
140 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
141
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
142 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
143 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
144 else {
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
145 print STDERR "incompatible subject and query alignement length\n $query_aln\n$subject_aln\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
146 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
147
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
148 # if ($line=~/chrCnn_random_49828229/){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
149 # print $line;
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
150 # print $query_aln,"\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
151 # print $subject_aln,"\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
152 # print $compt_mismatch_5p,"\t",$compt_mismatch_3p,"\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
153 # print $hash_name{"chrCnn_random_49828229"},"\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
154 # print "\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
155 # }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
156
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
157
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
158
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
159 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
160 # exit(0);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
161
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
162 close (INB);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
163
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
164 open(INV, $input_variant_file) or die ("Can't open $input_variant_file\n");
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
165
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
166 while (my $ligne = <INV>) {
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
167 $nb_variant_checked++;
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
168
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
169 my @champs = split (/\s+/,$ligne);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
170 my $header = $champs[0]."_".$champs[1];
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
171
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
172 if ($hash_name{$header}){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
173 if ($hash_name{$header}==1){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
174 print $ligne;
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
175 $nb_variant_selected++;
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
176 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
177 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
178 else {
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
179 #print STDERR "No blast result for ",$header,"\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
180 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
181
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
182
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
183 }
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
184
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
185 close(INV);
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
186
7
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
187 open (LF,">$log_file") or die("Can't open $log_file\n");
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
188 print LF "\n####\t Blast filtering \n";
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
189 print LF "Variant checked :\t$nb_variant_checked\n";
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
190 print LF "Variant selected :\t$nb_variant_selected\n";
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
191 close (LF);
3f7b0788a1c4 Uploaded
mcharles
parents: 0
diff changeset
192
0
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
193
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
194 # foreach my $key (sort keys %hash_name){
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
195 # print $key,"\t",$hash_name{$key},"\n";
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
196
442a7c88b886 Uploaded
mcharles
parents:
diff changeset
197 # }