annotate bismark_methyl_extractor/bismark2bedGraph @ 7:fcadce4d9a06 draft

planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
author bgruening
date Sat, 06 May 2017 13:18:09 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
7
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
1 #!/usr/bin/env perl
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
2 use warnings;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
3 use strict;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
4 $|++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
5 use Getopt::Long;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
6 use Cwd;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
7 use Carp;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
8
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
9 ## This program is Copyright (C) 2010-16, Felix Krueger (felix.krueger@babraham.ac.uk)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
10
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
11 ## This program is free software: you can redistribute it and/or modify
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
12 ## it under the terms of the GNU General Public License as published by
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
13 ## the Free Software Foundation, either version 3 of the License, or
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
14 ## (at your option) any later version.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
15
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
16 ## This program is distributed in the hope that it will be useful,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
17 ## but WITHOUT ANY WARRANTY; without even the implied warranty of
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
18 ## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
19 ## GNU General Public License for more details.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
20
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
21 ## You should have received a copy of the GNU General Public License
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
22 ## along with this program. If not, see <http://www.gnu.org/licenses/>.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
23
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
24 my $bismark2bedGraph_version = 'v0.16.3';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
25
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
26 my @bedfiles;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
27 my @methylcalls = qw (0 0 0); # [0] = methylated, [1] = unmethylated, [2] = total
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
28 my @sorting_files;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
29
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
30 my ($bedGraph_output,$parent_dir,$output_dir,$remove,$CX_context,$no_header,$sort_size,$coverage_threshold,$counts,$gazillion,$ample_mem,$zero,$input_dir) = process_commandline();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
31
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
32 warn "Using these input files: @sorting_files\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
33 warn "\nSummary of parameters for bismark2bedGraph conversion:\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
34 warn '='x54,"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
35 warn "bedGraph output:\t\t$bedGraph_output\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
36 warn "output directory:\t\t>$output_dir<\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
37 if ($remove){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
38 warn "remove whitespaces:\t\tyes\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
39 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
40 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
41 warn "remove whitespaces:\t\tno\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
42 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
43 if ($CX_context){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
44 warn "CX context:\t\t\tyes\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
45 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
46 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
47 warn "CX context:\t\t\tno (CpG context only, default)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
48 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
49 if ($no_header){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
50 warn "No-header selected:\t\tyes\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
51 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
52 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
53 warn "No-header selected:\t\tno\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
54 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
55
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
56 if ($ample_mem){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
57 warn "Sorting method:\t\t\tArray-based (faster, but larger memory footprint)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
58 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
59 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
60 warn "Sorting method:\t\t\tUnix sort-based (smaller memory footprint, but slower)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
61 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
62 unless($ample_mem){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
63 warn "Sort buffer size:\t\t$sort_size\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
64 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
65 warn "Coverage threshold:\t\t$coverage_threshold\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
66
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
67
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
68 warn "="x77,"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
69 warn "Methylation information will now be written into a bedGraph and coverage file\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
70 warn "="x77,"\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
71 sleep (2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
72
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
73 ### deciding which files to use for bedGraph conversion
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
74 foreach my $filename (@sorting_files){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
75
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
76 ### DETERMINING THE FULL PATH OF INPUT FILES
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
77 if ($filename =~ /^(.*\/)(.*)$/){ # if files are in a different output folder we extract the filename again
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
78 # warn "folder name: $1\nfilename: $2\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
79 chdir $1 or die "Failed to change directory to $1\n"; # $1 might be a relative path
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
80 $input_dir = getcwd(); # this will always be the full path
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
81 $filename = $2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
82 # warn "Full Input folder: $input_dir\nFilename: $filename\n\n"; sleep (1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
83 chdir $parent_dir or die "Failed to move back to the parent directory\n\n"; # moving back
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
84 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
85 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
86 $input_dir = $parent_dir;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
87 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
88 $input_dir .= '/';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
89
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
90 if ($CX_context){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
91 # push @bedfiles,$output_dir.$filename;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
92 push @bedfiles,$input_dir.$filename;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
93 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
94 else{ ## CpG context only (default)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
95 if ($filename =~ /^CpG/){ # only testing the actual filename without the path information
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
96 push @bedfiles,$input_dir.$filename; # we are adding the full path to the filename
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
97 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
98 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
99 # skipping CHH or CHG files
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
100 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
101 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
102 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
103
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
104 if (@bedfiles){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
105 warn "Using the following files as Input:\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
106 print join ("\t",@bedfiles),"\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
107 sleep (2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
108 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
109 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
110 die "It seems that you are trying to generate bedGraph files for files not starting with CpG.... Please specify the option '--CX' and try again\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
111 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
112
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
113 open (OUT,"| gzip -c - > ${output_dir}${bedGraph_output}") or die "Problems with the bedGraph output filename detected: file path: '$output_dir'\tfile name: '$bedGraph_output' $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
114 warn "Writing bedGraph to file: $bedGraph_output\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
115 print OUT "track type=bedGraph\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
116
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
117 my $coverage_output = $bedGraph_output;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
118 unless ($coverage_output =~ s/bedGraph\.gz$/bismark.cov.gz/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
119 $coverage_output =~ s/$/.bismark.cov.gz/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
120 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
121
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
122 open (COVERAGE,"| gzip -c - > ${output_dir}${coverage_output}") or die "Problems writing to the coverage output detected. File path: '$output_dir'\tfile name: '$coverage_output' $!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
123 warn "Also writing out a coverage file including counts methylated and unmethylated residues to file: $coverage_output\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
124
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
125 if ($zero){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
126 my $zero_coverage_output = $bedGraph_output;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
127 unless ($zero_coverage_output =~ s/bedGraph$/bismark.zero.cov/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
128 $zero_coverage_output =~ s/$/.bismark.zero.cov/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
129 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
130
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
131 open (ZEROCOVERAGE,'>',$output_dir.$zero_coverage_output) or die "Problems writing to the zero-based coverage output detected. File path: '$output_dir'\tfile name: '$\zero_coverage_output' $!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
132 warn "Also writing out a 0-based, half-open coverage file including counts methylated and unmethylated residues to file: $zero_coverage_output\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
133 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
134 warn "\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
135
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
136 my %temp_fhs;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
137 my @temp_files; # writing all context files (default CpG only) to these files prior to sorting
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
138 my %chr_lengths; # storing chromosome lenghts in '--ample_memory' mode
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
139
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
140 ### changing to the output directory
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
141 unless ($output_dir eq ''){ # default
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
142 chdir $output_dir or die "Failed to change directory to $output_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
143 warn "Changed directory to $output_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
144 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
145
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
146 if ($gazillion){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
147 if (scalar @bedfiles == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
148 warn "The genome of interest was specified to contain gazillions of chromosomes or scaffolds. Sorting everything in memory instead of writing out individual chromosome files ...\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
149 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
150 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
151 warn "The genome of interest was specified to contain gazillions of chromosomes or scaffolds. Merging all input files and sorting everything in memory instead of writing out individual chromosome files...\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
152 my $merge = "$bedGraph_output.methylation_calls.merged";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
153 open (MERGE,'>',$merge) or die "Failed to write to temporary merged file $merge: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
154 warn "Writing all merged methylation calls to temp file $merge\n\n"; sleep(2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
155 push @temp_files, $merge;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
156 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
157 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
158
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
159 foreach my $infile (@bedfiles) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
160
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
161 # warn "Processing $infile\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
162
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
163 if ($remove) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
164 warn "Now replacing whitespaces in the sequence ID field of the Bismark methylation extractor output $infile prior to bedGraph conversion\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
165
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
166 if ($infile =~ /gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
167 open (READ,"gunzip -c $infile |") or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
168 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
169 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
170 open (READ,$infile) or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
171 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
172
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
173 my $removed_spaces_outfile = $infile;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
174 $removed_spaces_outfile =~ s/$/.spaces_removed.txt/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
175 $removed_spaces_outfile =~ s/.*\///;# replacing everything up to the last slash in the filename
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
176
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
177 warn "Attempting to write to file ${output_dir}${removed_spaces_outfile}\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
178 open (REM,'>',$output_dir.$removed_spaces_outfile) or die "Couldn't write to file $removed_spaces_outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
179
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
180 unless ($no_header){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
181 $_ = <READ>; ### Bismark version header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
182 print REM $_; ### Bismark version header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
183 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
184
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
185 while (<READ>) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
186 chomp;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
187 my ($id,$strand,$chr,$pos,$context) = (split (/\t/));
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
188 $id =~ s/\s+/_/g;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
189 print REM join ("\t",$id,$strand,$chr,$pos,$context),"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
190 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
191
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
192 close READ or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
193 close REM or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
194
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
195 ### changing the infile name to the new file without spaces
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
196 $infile = $removed_spaces_outfile; # at this stage we are already in the output directory so it should pick it up correctly
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
197 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
198
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
199 # opening infile
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
200 if ($infile =~ /gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
201 open (IN,"gunzip -c $infile |") or die "Couldn't find file '$infile': $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
202 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
203 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
204 open (IN,$infile) or die "Couldn't find file '$infile': $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
205 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
206
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
207 if ($infile =~ /\//){ # if input files are in a different folders we extract the filename again
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
208 $infile =~ s/.*\///;# replacing everything up to the last slash in the filename
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
209 # warn "Renamed Infile: $infile\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
210 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
211
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
212 ### People these days seem to be aligning their data to newly assembled genomes more and more, which sometimes conist of up to half a million scaffolds instead of ~23 chromosomes. This
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
213 ### does normally clash with the operating system's limit of files that can be open for writing at the same time, and it is difficult and probably not advisable to increase this
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
214 ### limit (some even say there is a reason for the OS doing so...).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
215 ### To still allow then generation of bedGraph files we will in these cases sort everything using the Linux sort command instead, which will sort by chromosome and position (the
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
216 ### chromosome sorting is not carried out for chromosome sorted files which makes the sort MUCH faster).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
217
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
218 if ($gazillion){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
219 # using all infiles instead of sorting
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
220 if (scalar @bedfiles == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
221 push @temp_files, $infile;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
222 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
223 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
224 ## always ignoring the version header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
225 unless ($no_header){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
226 $_ = <IN>; ### Bismark version header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
227 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
228
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
229 while (<IN>) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
230 if ($_ =~ /^Bismark /){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
231 warn "Found Bismark version information. Skipping this line (should still work fine) but consider using '--no_header' next time...\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
232 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
233 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
234 print MERGE;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
235 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
236 warn "Finished writing methylation calls from $infile to merged temp file\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
237 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
238 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
239 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
240 warn "Now writing methylation information for file >>$infile<< to individual files for each chromosome\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
241
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
242 ## always ignoring the version header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
243 unless ($no_header){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
244 $_ = <IN>; ### Bismark version header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
245 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
246
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
247 while (<IN>) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
248 if ($_ =~ /^Bismark /){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
249 warn "Found Bismark version information. Skipping this line (should still work fine) but consider losing '--no_header' next time...\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
250 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
251 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
252
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
253 chomp;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
254
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
255 my ($chr,$pos) = (split (/\t/))[2,3];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
256
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
257 ### If --ample_mem was specified we are keeping track of the highest position for each chromosome as this will determine the size of the array we need to create in the next step
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
258 if ($ample_mem){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
259 ### setting the first position for this chromosome
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
260 unless (defined $chr_lengths{$chr} ){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
261 $chr_lengths{$chr} = $pos;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
262 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
263 # for all subsequent postions for this chromosome
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
264 if ($pos > $chr_lengths{$chr} ){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
265 $chr_lengths{$chr} = $pos; # set the current position as the new highest position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
266 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
267 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
268
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
269 # warn "This is the chromosome name before replacing '|' characters:\t$chr\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
270 $chr =~ s/\|/_/g; # replacing pipe ('|') characters in the file names
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
271 # warn "This is the chromosome name AFTER replacing '|' characters:\t$chr\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
272 unless (exists $temp_fhs{$chr}) { # Including the infile name to the temporary chromosome files to enable parallel processing of multiple files at the same time
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
273
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
274 my $temp_file_name = $infile.'.chr'.$chr.'.methXtractor.temp';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
275 # warn "Using temp file name: $temp_file_name\n"; sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
276
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
277 open ($temp_fhs{$chr},'>',$infile.'.chr'.$chr.'.methXtractor.temp') or die "Failed to open filehandle: $!";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
278 push @temp_files, $temp_file_name; # storing temp files as we open them instead
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
279 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
280
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
281 print {$temp_fhs{$chr}} "$_\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
282 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
283
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
284 warn "Finished writing out individual chromosome files for $infile\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
285 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
286 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
287
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
288 # closing temporary filehandles to force writing out buffered content
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
289 foreach my $temp_fh(keys %temp_fhs){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
290 close $temp_fhs{$temp_fh} or warn "Failed to close temporary filehandle $temp_fhs{$temp_fh}: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
291 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
292
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
293 ### printing out the determined maximum position for each chromosome
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
294 if ($ample_mem){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
295 foreach my $chr (sort keys %chr_lengths){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
296 warn "Highest determined position for chromosome $chr:\t\t$chr_lengths{$chr} bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
297 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
298 warn "\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
299 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
300
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
301 unless ($gazillion){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
302 warn "\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
303 warn "Collecting temporary chromosome file information... Processing the following input file(s):\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
304 warn join ("\n",@temp_files),"\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
305 sleep (1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
306 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
307
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
308 if ($gazillion){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
309 if (scalar @bedfiles > 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
310 close (MERGE) or die "Failed to close filehandle MERGE: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
311 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
312 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
313
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
314 foreach my $in (@temp_files) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
315
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
316 if ($sort_size){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
317 warn "Sorting input file $in by positions (using -S of $sort_size)\n" unless ($ample_mem);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
318 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
319
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
320 my $ifh;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
321
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
322 my $name;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
323 my $meth_state;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
324 my $chr = "";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
325 my $pos = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
326 my $meth_state2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
327
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
328 my $last_pos;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
329 my $last_chr;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
330
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
331 ### If the user specified to have a lot of RAM available (probably in the range of > 16GB for 2 arrays of human genome Chromosome 1) we will sort the methylation calls in two big arrays instead of using the Unix sort command
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
332 if ($ample_mem){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
333 # warn "Generating enormous array instead of sorting the file. This may temporily use quite a bit of memory (RAM)!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
334
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
335 my @meth_count;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
336 my @unmeth_count;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
337
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
338 open ($ifh,$in) or die "Couldn't read from temporary file '$in': $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
339
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
340 while (my $line = <$ifh>){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
341 next if ($line =~ /^Bismark/);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
342 chomp $line;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
343
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
344 ($name, $meth_state, $chr, $pos, $meth_state2) = split "\t", $line;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
345
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
346 unless ($last_pos and $last_chr){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
347 $last_chr = $chr;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
348 $last_pos = $pos;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
349 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
350 unless (@meth_count and @unmeth_count){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
351 warn "Setting maximum position of arrays \@meth_count and \@unmeth_count for chromosome $chr to $chr_lengths{$chr}\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
352 @meth_count = (0) x $chr_lengths{$chr};
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
353 @unmeth_count = (0) x $chr_lengths{$chr};
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
354 # warn "length of array meth count: ",scalar @meth_count,"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
355 warn "Finished generating arrays\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
356 # sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
357 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
358 # warn "Chromosome\tStart Position\tEnd Position\tMethylation Percentage\n"; sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
359 # print join ("\t",$name, $meth_state, $chr, $pos, $meth_state2),"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
360 # sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
361
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
362 # if ($last_chr ne $chr) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
363 # die "Reached new chromosome '$chr' which mustn't happen from pre-sorted files (previous chromosome was: '$last_chr')\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
364 # }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
365
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
366 my $validated = validate_methylation_call($meth_state, $meth_state2); # as a comment, methylation calls in Unknown context (U, u) would fail this check, but they should be ignored by the methylation extractor anyway
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
367 unless($validated){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
368 warn "Methylation state of sequence ($name) in file ($in) on line $. is inconsistent (meth_state is $meth_state, meth_state2 = $meth_state2)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
369 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
370 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
371
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
372 if ($meth_state eq '+'){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
373 # warn "increasing meth $pos by 1\n"; sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
374 $meth_count[$pos-1]++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
375 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
376 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
377 $unmeth_count[$pos-1]++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
378 # warn "increasing unmeth $pos by 1\n"; sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
379 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
380 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
381
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
382 close $ifh or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
383
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
384 warn "Now printing methylation information for this chromosome\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
385 # warn "length of array meth count: ",scalar @meth_count,"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
386 # warn "chr\tposition\tcount methylated\tcount unmethylated\tcount total\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
387 foreach my $index (0..$#meth_count){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
388 my $totalcount = $meth_count[$index] + $unmeth_count[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
389 if ($totalcount > 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
390 # warn "$index\t$meth_count[$index]\t$unmeth_count[$index]\t$totalcount\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
391 # sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
392
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
393 my $bed_pos = $index; ### bedGraph coordinates are 0 based
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
394 my $one_based_pos = $bed_pos + 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
395
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
396 my $meth_percentage;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
397 ($totalcount >= $coverage_threshold) ? ($meth_percentage = ($meth_count[$index]/$totalcount) * 100) : ($meth_percentage = undef);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
398
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
399 if (defined $meth_percentage){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
400
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
401 # as of version 0.9.1 we will by default write out both a bedGraph and a more detailed coverage file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
402
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
403 # this is the bedGraph file, the starting position is 0-based, the end position is 1-based! (half-open. Clever, huh?)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
404 print OUT "$last_chr\t$bed_pos\t$one_based_pos\t$meth_percentage\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
405
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
406 # this is the coverage file. Coordinates are 1-based
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
407 print COVERAGE "$last_chr\t$one_based_pos\t$one_based_pos\t$meth_percentage\t$meth_count[$index]\t$unmeth_count[$index]\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
408
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
409 # this is an optional 0-based, half-open coverage file. Coordinates are 0-based start and 1-based end
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
410 if ($zero){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
411 print ZEROCOVERAGE "$last_chr\t$bed_pos\t$one_based_pos\t$meth_percentage\t$meth_count[$index]\t$unmeth_count[$index]\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
412 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
413
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
414 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
415 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
416 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
417
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
418 @meth_count = ();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
419 @unmeth_count = ();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
420
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
421 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
422 ### default: we assume that the user wants to use the Linux Sort command. This is quite a bit slower, but features a much smaller memory footprint
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
423 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
424 my $sort_dir = './'; # there has been a cd into the output_directory already
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
425 # my $sort_dir = $output_dir;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
426 # if ($sort_dir eq ''){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
427 # $sort_dir = './';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
428 # }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
429
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
430 if ($gazillion){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
431 if ($in =~ /gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
432 open $ifh, "gunzip -c $in | sort -S $sort_size -T $sort_dir -k3,3V -k4,4n |" or die "Input file could not be sorted. $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
433 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
434 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
435 open $ifh, "sort -S $sort_size -T $sort_dir -k3,3V -k4,4n $in |" or die "Input file could not be sorted. $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
436 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
437 ### Comment by Volker Brendel, Indiana University
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
438 ### "The -k3,3V sort option is critical when the sequence names are numbered scaffolds (without left-buffering of zeros). Omit the V, and things go very wrong in the tallying of reads."
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
439 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
440 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
441 ### this sort command was used previously and sorts according to chromosome in addition to position. Since the files are being sorted according to chromosomes anyway,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
442 ### we may drop the -k3,3V option. It has been reported that this will result in a dramatic speed increase
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
443 if ($in =~ /gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
444 open $ifh, "gunzip -c $in | sort -S $sort_size -T $sort_dir -k4,4n |" or die "Input file could not be sorted. $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
445 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
446 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
447 open $ifh, "sort -S $sort_size -T $sort_dir -k4,4n $in |" or die "Input file could not be sorted. $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
448 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
449 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
450
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
451 while (my $line = <$ifh>) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
452 next if ($line =~ /^Bismark/);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
453 chomp $line;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
454 #warn "full line:\n$line\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
455 $last_chr = $chr;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
456 $last_pos = $pos;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
457 ($name, $meth_state, $chr, $pos, $meth_state2) = split "\t", $line;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
458 #warn "$name, $meth_state, $chr, $pos, $meth_state2\n"; sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
459 if (($last_pos ne $pos) || ($last_chr ne $chr)) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
460 generate_output($last_chr,$last_pos) if $methylcalls[2] > 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
461 @methylcalls = qw (0 0 0);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
462 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
463
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
464 my $validated = validate_methylation_call($meth_state, $meth_state2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
465 unless($validated){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
466 warn "Methylation state of sequence ($name) in file ($in) on line $. is inconsistent (meth_state is $meth_state, meth_state2 = $meth_state2)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
467 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
468 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
469 if ($meth_state eq "+") {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
470 $methylcalls[0]++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
471 $methylcalls[2]++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
472 } else {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
473 $methylcalls[1]++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
474 $methylcalls[2]++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
475 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
476 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
477
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
478
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
479 $last_chr = $chr;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
480 $last_pos = $pos;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
481 if ($methylcalls[2] > 0) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
482 generate_output($last_chr,$last_pos) if $methylcalls[2] > 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
483 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
484
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
485 close $ifh or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
486
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
487 @methylcalls = qw (0 0 0); # resetting @methylcalls
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
488
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
489 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
490
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
491 ### deleting temporary files (only needed if --gazillion hasn't been specified
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
492 if ($gazillion and scalar @bedfiles == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
493 # if there was only 1 file to sort this will be the input file, which obviously shouldn't be removed
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
494 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
495 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
496 my $delete = unlink $in;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
497 if ($delete) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
498 warn "Successfully deleted the temporary input file $in\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
499 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
500 else {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
501 warn "The temporary inputfile $in could not be deleted $!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
502 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
503 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
504 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
505
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
506 close OUT or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
507 close COVERAGE or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
508 if ($zero){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
509 close ZEROCOVERAGE or die $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
510 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
511
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
512 exit 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
513
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
514
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
515
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
516 sub validate_methylation_call{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
517 my $meth_state = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
518 croak "Missing (+/-) methylation call" unless defined $meth_state;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
519 my $meth_state2 = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
520 croak "Missing alphabetical methylation call" unless defined $meth_state2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
521 my $is_consistent;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
522 ($meth_state2 =~ /^z/i) ? ($is_consistent = check_CpG_methylation_call($meth_state, $meth_state2))
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
523 : ($is_consistent = check_nonCpG_methylation_call($meth_state,$meth_state2));
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
524 return 1 if $is_consistent;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
525 return 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
526 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
527
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
528 sub check_CpG_methylation_call{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
529 my $meth1 = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
530 my $meth2 = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
531 return 1 if($meth1 eq "+" && $meth2 eq "Z");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
532 return 1 if($meth1 eq "-" && $meth2 eq "z");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
533 return 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
534 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
535
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
536 sub check_nonCpG_methylation_call{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
537 my $meth1 = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
538 my $meth2 = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
539 return 1 if($meth1 eq "+" && $meth2 eq "C");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
540 return 1 if($meth1 eq "+" && $meth2 eq "X");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
541 return 1 if($meth1 eq "+" && $meth2 eq "H");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
542 return 1 if($meth1 eq "-" && $meth2 eq "c");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
543 return 1 if($meth1 eq "-" && $meth2 eq "x");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
544 return 1 if($meth1 eq "-" && $meth2 eq "h");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
545 return 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
546 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
547
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
548 sub generate_output{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
549 my $methcount = $methylcalls[0];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
550 my $nonmethcount = $methylcalls[1];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
551 my $totalcount = $methylcalls[2];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
552 my $last_chr = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
553 my $last_pos = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
554 croak "Should not be generating output if there's no reads to this region" unless ($totalcount > 0);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
555 croak "Total counts ($totalcount) is not the sum of the methylated ($methcount) and unmethylated ($nonmethcount) counts" if ($totalcount != ($methcount + $nonmethcount) );
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
556
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
557 my $bed_pos = $last_pos - 1; ### Bismark coordinates are 1 based whereas bedGraph coordinates are 0 based.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
558 my $meth_percentage;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
559 ($totalcount >= $coverage_threshold) ? ($meth_percentage = ($methcount/$totalcount) * 100) : ($meth_percentage = undef);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
560 # $meth_percentage =~ s/(\.\d\d).+$/$1/ unless $meth_percentage =~ /^Below/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
561 if (defined $meth_percentage){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
562
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
563 # this is the bedGraph file, the starting position is 0-based, the end position is 1-based! (clever, huh?)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
564 my $one_based_pos = $bed_pos + 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
565 print OUT "$last_chr\t$bed_pos\t$one_based_pos\t$meth_percentage\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
566
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
567 # this is the coverage file. Coordinates are 1-based
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
568 print COVERAGE "$last_chr\t$one_based_pos\t$one_based_pos\t$meth_percentage\t$methcount\t$nonmethcount\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
569
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
570 # this is an optional 0-based, half-open coverage file. Coordinates are 0-based start and 1-based end
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
571 if ($zero){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
572 print ZEROCOVERAGE "$last_chr\t$bed_pos\t$one_based_pos\t$meth_percentage\t$methcount\t$nonmethcount\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
573 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
574 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
575
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
576 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
577
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
578 sub process_commandline{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
579 my $help;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
580 my $output_dir;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
581 my $bedGraph_output;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
582 my $no_header;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
583 my $coverage_threshold; # Minimum number of reads covering before calling methylation status
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
584 my $remove;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
585 my $counts;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
586 my $CX_context;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
587 my $sort_size;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
588 my $version;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
589 my $gazillion;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
590 my $ample_mem;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
591 my $zero;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
592 my $input_dir;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
593
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
594 my $command_line = GetOptions ('help|man' => \$help,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
595 'dir=s' => \$output_dir,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
596 'o|output=s' => \$bedGraph_output,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
597 'no_header' => \$no_header,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
598 "cutoff=i" => \$coverage_threshold,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
599 "remove_spaces" => \$remove,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
600 "counts" => \$counts,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
601 "CX|CX_context" => \$CX_context,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
602 "buffer_size=s" => \$sort_size,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
603 'version' => \$version,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
604 'gazillion|scaffolds' => \$gazillion,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
605 'ample_memory' => \$ample_mem,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
606 "zero_based" => \$zero,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
607 );
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
608
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
609 ### EXIT ON ERROR if there were errors with any of the supplied options
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
610 unless ($command_line){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
611 die "Please respecify command line options\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
612 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
613
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
614 ### HELPFILE
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
615 if ($help){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
616 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
617 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
618 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
619
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
620 if ($version){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
621 print << "VERSION";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
622
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
623
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
624 Bismark Methylation Extractor Module -
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
625 bismark2bedGraph
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
626
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
627 Bismark Extractor Version: $bismark2bedGraph_version
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
628 Copyright 2010-15 Felix Krueger, Babraham Bioinformatics
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
629 www.bioinformatics.babraham.ac.uk/projects/bismark/
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
630
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
631
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
632 VERSION
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
633 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
634 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
635
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
636 @sorting_files = @ARGV;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
637
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
638 ### no files provided
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
639 unless (@sorting_files){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
640 warn "You need to provide one or more Bismark methylation caller files to create an individual C methylation bedGraph output. Please respecify!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
641 sleep(2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
642
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
643 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
644 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
645 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
646
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
647 ### PARENT DIRECTORY
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
648 my $parent_dir = getcwd();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
649 # warn "parent directory is: $parent_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
650
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
651 ### OUTPUT DIR PATH
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
652 if (defined $output_dir){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
653
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
654 unless ($output_dir eq ''){ # if the output dir has been passed on by the methylation extractor and is an empty string we don't want to change it
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
655 unless ($output_dir =~ /\/$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
656 $output_dir =~ s/$/\//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
657 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
658 unless (-d $output_dir){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
659 mkdir $output_dir or die "Failed to create output directory $output_dir: $!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
660 warn "Created output directory $output_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
661 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
662
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
663 ### want to get an absolute path for the output directory instead of a relative one
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
664 chdir $output_dir or die "Failed to move into output directory '$output_dir': $!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
665 $output_dir = getcwd();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
666 unless ($output_dir =~ /\/$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
667 $output_dir =~ s/$/\//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
668 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
669 # warn "output directory is: $output_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
670
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
671 # changing back to the parent directory
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
672 chdir $parent_dir or die "Failed to move back into parent directory '$parent_dir': $!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
673
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
674 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
675
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
676 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
677 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
678 $output_dir = '';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
679 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
680
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
681 unless (defined $bedGraph_output){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
682 die "Please provide the name of the output file using the option -o/--output filename\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
683 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
684
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
685 if ($bedGraph_output =~ /\//){ # this is supposed a filename and not a path name
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
686 die "Please specify a file name without any path information (or use --dir if necessary)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
687 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
688
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
689 unless ($bedGraph_output =~ /\.gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
690 $bedGraph_output = "${bedGraph_output}.gz"; ### 22 07 2015: Output will be gzip compressed
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
691 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
692
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
693 ### NO HEADER
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
694 unless ($no_header){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
695 $no_header = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
696 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
697
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
698 ### remove white spaces in read ID (needed for sorting using the sort command
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
699 unless ($remove){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
700 $remove = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
701 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
702
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
703 ### COVERAGE THRESHOLD FOR bedGraph OUTPUT
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
704 if (defined $coverage_threshold){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
705 unless ($coverage_threshold > 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
706 die "Please select a coverage greater than 0 (positive integers only)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
707 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
708 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
709 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
710 $coverage_threshold = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
711 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
712
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
713 ### SORT buffer size
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
714 if (defined $sort_size){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
715 unless ($sort_size =~ /^\d+\%$/ or $sort_size =~ /^\d+(K|M|G|T)$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
716 die "Please select a buffer size as percentage (e.g. --buffer_size 20%) or a number to be multiplied with K, M, G, T etc. (e.g. --buffer_size 20G). For more information on sort type 'info sort' on a command line\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
717 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
718 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
719 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
720 $sort_size = '2G';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
721 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
722
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
723 unless ($CX_context){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
724 $CX_context = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
725 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
726
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
727 unless ($counts){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
728 $counts = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
729 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
730
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
731 if ($gazillion){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
732 if ($ample_mem){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
733 die "You can't currently select '--ample_mem' together with '--gazillion'. Make your pick!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
734 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
735 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
736
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
737 return ($bedGraph_output,$parent_dir,$output_dir,$remove,$CX_context,$no_header,$sort_size,$coverage_threshold,$counts,$gazillion,$ample_mem,$zero,$input_dir);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
738 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
739
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
740
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
741 sub print_helpfile{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
742 print <<EOF
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
743
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
744 SYNOPSIS:
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
745
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
746 This script uses positional methylation data generated by the Bismark methylation extractor to generate
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
747 a bedGraph file as well as a coverage file which are both sorted by chromosomal position. The bedGraph
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
748 file uses 0-based genomic start and 1-based genomic end coordinates and should be UCSC compatible (if
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
749 UCSC genomes were used for the alignment step). In addition this module will write out a coverage file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
750 which is similar to the bedGraph file, but uses 1-based genomic coordinates and also reports the count
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
751 of methylated and unmethylated cytosines for any covered position; this coverage file is required if you
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
752 wish to generate a genome-wide cytosine report with the module coverage2cytosine.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
753
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
754 USAGE: bismark2bedGraph [options] -o <output> [methylation extractor input files]
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
755
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
756 Methylation extractor input files: These files are required to start with CpG... in order for the
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
757 script to correctly work out the sequence context when using CpG context only (default). If all cytosine
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
758 contexts are selected ('--CX_context'), all input files will be used regardless of their file file name(s).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
759
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
760
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
761 -o/--output <filename> Name of the output file, mandatory.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
762
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
763 --dir Output directory. Output is written to the current directory if not specified explicitly.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
764
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
765 --cutoff [threshold] The minimum number of times a methylation state has to be seen for that nucleotide
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
766 before its methylation percentage is reported. Default: 1.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
767
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
768 --remove_spaces Replaces whitespaces in the sequence ID field with underscores to allow sorting.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
769
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
770 --CX/--CX_context The sorted bedGraph output file contains information on every single cytosine that was covered
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
771 in the experiment irrespective of its sequence context. This applies to both forward and
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
772 reverse strands. Please be aware that this option may generate large temporary and output files
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
773 and may take a long time to sort (up to many hours). Default: OFF.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
774 (i.e. Default = CpG context only).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
775
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
776 --buffer_size <string> This allows you to specify the main memory sort buffer when sorting the methylation information.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
777 Either specify a percentage of physical memory by appending % (e.g. --buffer_size 50%) or
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
778 a multiple of 1024 bytes, e.g. 'K' multiplies by 1024, 'M' by 1048576 and so on for 'T' etc.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
779 (e.g. --buffer_size 20G). For more information on sort type 'info sort' on a command line.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
780 Defaults to 2G.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
781
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
782 --scaffolds/--gazillion Users working with unfinished genomes sporting tens or even hundreds of thousands of
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
783 scaffolds/contigs/chromosomes frequently encountered errors with pre-sorting reads to
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
784 individual chromosome files. These errors were caused by the operating system's limit
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
785 of the number of filehandle that can be written to at any one time (typically 1024; to
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
786 find out this limit on Linux, type: ulimit -a).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
787 To bypass the limitation of open filehandles, the option --scaffolds does not pre-sort
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
788 methylation calls into individual chromosome files. Instead, all input files are
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
789 temporarily merged into a single file (unless there is only a single file), and this
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
790 file will then be sorted by both chromosome AND position using the Unix sort command.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
791 Please be aware that this option might take a looooong time to complete, depending on
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
792 the size of the input files, and the memory you allocate to this process (see --buffer_size).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
793 Nevertheless, it seems to be working.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
794
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
795 --ample_memory Using this option will not sort chromosomal positions using the UNIX 'sort' command, but will
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
796 instead use two arrays to sort methylated and unmethylated calls, respectively. This may result
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
797 in a faster sorting process for very large files, but this comes at the cost of a larger memory
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
798 footprint (as an estimate, two arrays of the length of (the largest) human chromosome 1 (nearly
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
799 250 million bp) temporarily consume around 16GB of RAM). Note however that due to the overheads
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
800 of creating and looping through arrays this option might in fact be *slower* for small-ish
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
801 files (up to a few million alignments). Note also that this option is not currently compatible
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
802 with options '--scaffolds/--gazillion'.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
803
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
804 --zero_based Write out an additional coverage file (ending in .zero.cov) that uses 0-based genomic start
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
805 and 1-based genomic end coordinates (zero-based, half-open), like used in the bedGraph file,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
806 instead of using 1-based coordinates throughout. Default: OFF.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
807
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
808
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
809
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
810 The bedGraph output looks like this (tab-delimited; 0-based start coords, 1-based end coords):
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
811 ==============================================================================================
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
812
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
813 track type=bedGraph (header line)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
814
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
815 <chromosome> <start position> <end position> <methylation percentage>
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
816
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
817
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
818
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
819 The coverage output looks like this (tab-delimited, 1-based genomic coords; optional zero-based, half-open coords with '--zero_based'):
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
820 =======================================================================================================================================
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
821
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
822 <chromosome> <start position> <end position> <methylation percentage> <count methylated> <count non-methylated>
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
823
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
824
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
825 Script last modified: 09 December 2015
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
826
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
827 EOF
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
828 ;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
829 exit 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n'
bgruening
parents:
diff changeset
830 }