Mercurial > repos > brasset_jensen > srnapipe

annotate bin/sRNAPipe.pl @ 61:9185ca0a7b43 draft

Find changesets by keywords (author, files, the commit message), revision number or hash, or revset expression.
Updated package according to recommendations.
author pierre.pouchin
date Wed, 16 Jan 2019 08:18:13 -0500
parents 9645d995fb3c
children 967512924317
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
61
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
1 #!/usr/bin/perl
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
2 use strict;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
3 use warnings;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
4 use Getopt::Long;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
5 use Parallel::ForkManager;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
6 use File::Basename;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
7 use File::Copy::Recursive qw( dircopy );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
8 use POSIX;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
9 use FindBin;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
10 use lib $FindBin::Bin;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
11 use resize qw ( size_distribution );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
12 use subgroups qw (subgroups );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
13 use ppp qw ( ping_pong_partners );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
14 use Rcall qw (pie_chart bg_to_png );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
15 use align qw ( to_build get_unique sam_count sam_count_mis sam_sorted_bam rpms_rpkm rpms_rpkm_te BWA_call get_fastq_seq extract_sam sam_to_bam_bg );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
16 use html qw ( main_page details_pages menu_page ppp_page );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
17 use File::Copy;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
18
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
19 if(@ARGV) {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
20 my ( @fastq, @fastq_n, $dir, $min, $max, $mis, $misTE, $help, $Pcheck, $mapnumf, $html_out);
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
21 my ( $ref, $tRNAs, $rRNAs, $snRNAs, $miRNAs, $transcripts, $TE );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
22 my ( $si_min, $si_max, $pi_min, $pi_max );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
23 my ( $build_index, $build_tRNAs, $build_rRNAs, $build_snRNAs, $build_miRNAs, $build_transcripts, $build_TE );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
24 my $max_procs = 8;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
25
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
26 ( $build_index, $build_tRNAs, $build_rRNAs, $build_snRNAs, $build_miRNAs, $build_transcripts, $build_TE ) = (0,0,0,0,0,0,0);
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
27 ( $min, $max, $mis, $misTE, $si_min, $si_max, $pi_min, $pi_max, $dir ) = ( 18, 29, 0, 3, 21, 21, 23, 29 );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
28 $Pcheck ='true';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
29
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
30 GetOptions (
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
31 "fastq=s" => \@fastq,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
32 "fastq_n=s" => \@fastq_n,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
33 "dir=s" => \$dir,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
34 "min:i" => \$min,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
35 "max:i" => \$max,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
36 "si_min:i" => \$si_min,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
37 "si_max:i" => \$si_max,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
38 "pi_min:i" => \$pi_min,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
39 "pi_max:i" => \$pi_max,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
40 "mis:i" => \$mis,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
41 "misTE:i" => \$misTE,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
42 "html:s" => \$html_out,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
43 "PPPon:s" => \$Pcheck,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
44 "help" => \$help,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
45 "ref:s" => \$ref,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
46 "tRNAs:s" => \$tRNAs,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
47 "rRNAs:s" => \$rRNAs,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
48 "snRNAs:s" => \$snRNAs,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
49 "miRNAs:s" => \$miRNAs,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
50 "transcripts:s" => \$transcripts,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
51 "TE:s" => \$TE,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
52 "build_index" => \$build_index,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
53 "build_tRNAs" => \$build_tRNAs,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
54 "build_snRNAs" => \$build_snRNAs,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
55 "build_miRNAs" => \$build_miRNAs,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
56 "build_transcripts" => \$build_transcripts,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
57 "build_rRNAs" => \$build_rRNAs,
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
58 "build_TE" => \$build_TE
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
59 );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
60
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
61 my $fq_collection = 'fastq_dir/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
62 mkdir $dir; mkdir $fq_collection;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
63 $dir = $dir.'/' unless $dir =~ /\/$/;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
64 mkdir $dir.'/css';mkdir $dir.'/js';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
65 dircopy( $FindBin::Bin.'/css', $dir.'/css' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
66 dircopy( $FindBin::Bin.'/js', $dir.'/js' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
67
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
68 my $file = $dir.'report.txt';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
69 open my $report, '>', $file or die "Cannot open $file $!\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
70
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
71 my @toBuild = ( [$build_index, \$ref], [$build_tRNAs, \$tRNAs], [$build_rRNAs, \$rRNAs], [$build_snRNAs, \$snRNAs], [$build_miRNAs, \$miRNAs], [$build_transcripts, \$transcripts], [$build_TE, \$TE] );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
72 to_build ( \@toBuild, $report, $dir );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
73
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
74 my $proc_child = ceil($max_procs / scalar(@fastq));
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
75 my $proc_grand_child = ceil($proc_child/4);
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
76 my $pm = Parallel::ForkManager->new($max_procs);
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
77 my $pm2 = Parallel::ForkManager->new($proc_grand_child);
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
78
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
79 $pm->run_on_finish( sub {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
80 my ($pid, $exit_code, $ident) = @_;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
81 print $report "Fastq fork $ident just finished ".
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
82 "with PID $pid and exit code: $exit_code\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
83 die "Something went wrong!\n" if $exit_code != 0;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
84 });
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
85
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
86
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
87 $pm->run_on_start( sub {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
88 my ($pid,$ident)=@_;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
89 print $report "Fastq fork : $ident started, pid: $pid\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
90 });
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
91
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
92
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
93 $pm2->run_on_finish( sub {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
94 my ($pid, $exit_code, $ident) = @_;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
95 print $report "** Subgroup fork $ident just finished ".
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
96 "with PID $pid and exit code: $exit_code\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
97 die "Something went wrong!\n" if $exit_code != 0;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
98 });
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
99
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
100
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
101 $pm2->run_on_start( sub {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
102 my ($pid,$ident)=@_;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
103 print $report "** Subgroup fork $ident started, pid: $pid\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
104 });
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
105
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
106 foreach my $child ( 0 .. $#fastq )
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
107 {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
108 my @suffix = ('.fastq', '.fastq.gz,', '.fq', '.fq.gz', 'ref', '.dat', '.fa','.fas','.fasta', '.txt');
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
109 my ( $name, $path, $suffix ) = fileparse( $fastq[$child], @suffix );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
110 my ( $ref_name, $ref_path, $ref_suffix ) = fileparse( $ref, @suffix );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
111 my ( $TE_name, $TE_path, $TE_suffix ) = fileparse( $TE, @suffix );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
112 my ( $ex_name, $ex_path, $ex_suffix ) = fileparse( $transcripts, @suffix );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
113
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
114 $pm->start($fastq[$child]) and next;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
115
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
116 my $dir_fq = $dir.$fastq_n[$child].'/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
117 mkdir $dir_fq;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
118
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
119 my $gen_dir = $dir_fq.'genome/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
120 mkdir $gen_dir;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
121
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
122 my $size_dir = $dir_fq.'size/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
123 mkdir $size_dir;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
124
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
125 my $fastq_resized = $dir_fq.$name.'_'.$min.'-'.$max.'.fastq';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
126 size_distribution ( $fastq[$child], $fastq_resized, $size_dir, $min, $max );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
127
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
128 my $sam_genome = $gen_dir.$fastq_n[$child].'_'.$min.'-'.$max.'.sam';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
129 my $sam_genome_unique = $gen_dir.$fastq_n[$child].'_'.$min.'-'.$max.'_unique.sam';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
130 my $fastq_prefix = $gen_dir.$fastq_n[$child].'_'.$min.'-'.$max;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
131
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
132 BWA_call ( $ref, $fastq_resized, $sam_genome, $mis, $proc_child, $report );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
133 my ( $fai_ref_hashP, $ma, $ma_uni ) = get_unique ( $sam_genome, $sam_genome_unique, $gen_dir, $fq_collection.$fastq_n[$child], 1, $report );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
134
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
135 die "No Reads mapped on the genome reference!\n" if $ma == 0;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
136 my $scale = 1000000 / $ma;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
137 sam_to_bam_bg ( $sam_genome_unique, $scale, $proc_child );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
138 sam_to_bam_bg ( $sam_genome, $scale, $proc_child );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
139
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
140 my $Gviz_dir = $gen_dir.'Gviz/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
141 my $fai_file = $gen_dir.'fai';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
142 mkdir $Gviz_dir;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
143 my $Gviz_dir_rand = $Gviz_dir.'rand/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
144 mkdir $Gviz_dir_rand;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
145 my $Gviz_dir_uni = $Gviz_dir.'unique/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
146 mkdir $Gviz_dir_uni;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
147
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
148 open my $gfai, '>', $fai_file;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
149 foreach my $k ( sort keys %{$fai_ref_hashP} )
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
150 {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
151 print $gfai "$k\t$fai_ref_hashP->{$k}\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
152 }
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
153 close $gfai;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
154 bg_to_png ( $fai_file, $fastq_prefix.'_unique_plus.bedgraph', $fastq_prefix.'_unique_minus.bedgraph', $Gviz_dir_uni, 'Mb' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
155 bg_to_png ( $fai_file, $fastq_prefix.'_plus.bedgraph', $fastq_prefix.'_minus.bedgraph', $Gviz_dir_rand, 'Mb' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
156
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
157 my $group_dir = $dir_fq.'subgroups/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
158 my $fastq_uni = $fq_collection.$fastq_n[$child].'_unique_mappers.fastq';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
159 my $fastq_all = $fq_collection.$fastq_n[$child].'_all_mappers.fastq';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
160 my ($bo, $mi, $pi) = subgroups ( $fastq_all, $group_dir, $mis, $misTE, $proc_child, $tRNAs, $rRNAs, $snRNAs, $miRNAs, $transcripts, $TE, $si_min, $si_max, $pi_min, $pi_max, $report);
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
161
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
162 pie_chart($group_dir);
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
163
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
164 open (my $dupnum, $gen_dir.'dup_mapnum.txt') || die "cannot open dup_mapnum.txt $!";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
165 my %dupnum_genome;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
166 my $header = <$dupnum>;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
167 while (<$dupnum>)
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
168 {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
169 chomp $_;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
170 my @dupline = split /\t/, $_;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
171 $dupnum_genome{$dupline[0]} = [$dupline[1], $dupline[2]];
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
172 }
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
173 close $dupnum;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
174
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
175 my $mi_sam = $group_dir.'miRNAs.sam';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
176 mkdir $group_dir.'miRNAs/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
177 my $mi_count_file = $group_dir.'miRNAs/miRNAs_reads_counts.txt';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
178 my ( $mi_count, $mi_ref_size ) = sam_count ( $mi_sam );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
179
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
180 rpms_rpkm( $mi_count, $mi_ref_size, $ma, $mi_count_file, $pi, $mi, $bo );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
181
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
182 my ( $sam_transcripts, $sam_TEs ) = ( $group_dir.'transcripts.sam', $group_dir.'TEs.sam' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
183 my @types = ($group_dir.'bonafide_reads.fastq', $group_dir.'miRNAs.fastq', $group_dir.'siRNAs.fastq', $group_dir.'piRNAs.fastq' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
184 my @types_names = ('bonafide_reads', 'miRNAs', 'siRNAs', 'piRNAs');
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
185 foreach my $grand_child ( 0 .. $#types )
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
186 {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
187 my $type_dir = $group_dir.$types_names[$grand_child].'/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
188 my $type_prefix = $types_names[$grand_child].'-';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
189 mkdir $type_dir;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
190 $pm2->start($types[$grand_child]) and next;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
191 my ( $type_sam_genome, $type_sam_TEs, $type_sam_transcripts ) = ( $type_dir.$type_prefix.'genome.sam', $type_dir.$type_prefix.'TEs.sam', $type_dir.$type_prefix.'transcripts.sam' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
192 my ( $type_sam_uni_genome, $type_sam_uni_TEs, $type_sam_uni_transcripts ) = ( $type_dir.$type_prefix.'genome_unique.sam', $type_dir.$type_prefix.'TEs_unique.sam', $type_dir.$type_prefix.'transcripts_unique.sam' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
193 my ( $type_uni_genome_fastq, $type_uni_TEs_fastq, $type_uni_transcripts_fastq ) = ( $fq_collection.$fastq_n[$child].'-'.$type_prefix.'genome_uni.fastq', $fq_collection.$fastq_n[$child].'-'.$type_prefix.'TEs_uni.fastq', $fq_collection.$fastq_n[$child].'-'.$type_prefix.'transcripts_uni.fastq');
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
194 my ( $type_genome_fastq, $type_TEs_fastq, $type_transcripts_fastq ) = ( $fq_collection.$fastq_n[$child].'-'.$type_prefix.'genome.fastq', $fq_collection.$fastq_n[$child].'-'.$type_prefix.'TEs.fastq', $fq_collection.$fastq_n[$child].'-'.$type_prefix.'transcripts.fastq');
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
195 my $type_sequence_hashP = get_fastq_seq ( $types[$grand_child] );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
196
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
197 if ( $grand_child == 1 )
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
198 {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
199 BWA_call ( $TE, $types[$grand_child], $type_sam_TEs, $misTE, $proc_child, $report );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
200 BWA_call ( $transcripts, $types[$grand_child], $type_sam_transcripts, $mis, $proc_child, $report );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
201 BWA_call ( $ref, $types[$grand_child], $type_sam_genome, $mis, $proc_child, $report );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
202 extract_sam ( undef, $type_sam_TEs, $type_sam_TEs, $type_sam_uni_TEs, $type_uni_TEs_fastq, $type_uni_TEs_fastq );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
203 extract_sam ( undef, $type_sam_transcripts, $type_sam_transcripts, $type_sam_uni_transcripts, $type_transcripts_fastq, $type_uni_transcripts_fastq );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
204 extract_sam ( undef, $type_sam_genome, $type_sam_genome, $type_sam_uni_genome, $type_genome_fastq, $type_uni_genome_fastq );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
205 }
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
206 else
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
207 {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
208 extract_sam ( $type_sequence_hashP, $sam_TEs, $type_sam_TEs, $type_sam_uni_TEs, $type_TEs_fastq, $type_uni_TEs_fastq );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
209 extract_sam ( $type_sequence_hashP, $sam_transcripts, $type_sam_transcripts, $type_sam_uni_transcripts, $type_transcripts_fastq, $type_uni_transcripts_fastq );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
210 extract_sam ( $type_sequence_hashP, $sam_genome, $type_sam_genome, $type_sam_uni_genome, $type_genome_fastq, $type_uni_genome_fastq );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
211 }
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
212
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
213 my $ex_count_file = $type_dir.$type_prefix.'transcripts_reads_counts.txt';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
214 my ( $ex_count, $ex_ref_size ) = sam_count_mis ( $type_sam_transcripts );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
215 rpms_rpkm_te( $ex_count, $ex_ref_size, $ma, $ex_count_file, $pi, $mi, $bo );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
216
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
217 my ( $TEs_count, $TEs_ref_size, $TEs_count_NoM, $TEs_count_M ) = sam_count_mis ( $type_sam_TEs );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
218 my $TEs_count_file = $type_dir.$type_prefix.'TEs_reads_counts.txt';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
219 my $TEs_count_file_M = $type_dir.$type_prefix.'TEs_reads_counts_mismatches.txt';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
220 my $TEs_count_file_noM = $type_dir.$type_prefix.'TEs_reads_counts_nomismatches.txt';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
221 rpms_rpkm_te( $TEs_count, $TEs_ref_size, $ma, $TEs_count_file, $pi, $mi, $bo );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
222 rpms_rpkm_te( $TEs_count_NoM, $TEs_ref_size, $ma, $TEs_count_file_noM, $pi, $mi, $bo );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
223 rpms_rpkm_te( $TEs_count_M, $TEs_ref_size, $ma, $TEs_count_file_M, $pi, $mi, $bo );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
224
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
225 sam_to_bam_bg ( $type_sam_TEs, $scale, $grand_child );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
226 sam_sorted_bam ( $type_sam_transcripts, $grand_child ); sam_sorted_bam ( $type_sam_uni_transcripts, $grand_child );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
227 sam_sorted_bam ( $type_sam_uni_TEs, $grand_child );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
228
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
229 my $Gviz_TEs = $type_dir.'Gviz_TEs/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
230 mkdir $Gviz_TEs;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
231 bg_to_png ( $group_dir.'TEs.fai', $type_dir.$type_prefix.'TEs_plus.bedgraph', $type_dir.$type_prefix.'TEs_minus.bedgraph', $Gviz_TEs, 'Kb' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
232
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
233 my $Gviz_genome= $type_dir.'Gviz_genome/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
234 my $Gviz_genome_rand = $Gviz_genome.'rand/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
235 my $Gviz_genome_uni = $Gviz_genome.'unique/';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
236 mkdir $Gviz_genome; mkdir $Gviz_genome_uni; mkdir $Gviz_genome_rand;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
237
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
238 sam_to_bam_bg ( $type_sam_genome, $scale, $grand_child );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
239 sam_to_bam_bg ( $type_sam_uni_genome, $scale, $grand_child );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
240
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
241 bg_to_png ( $fai_file, $type_dir.$type_prefix.'genome_unique_plus.bedgraph', $type_dir.$type_prefix.'genome_unique_minus.bedgraph', $Gviz_genome_uni, 'Mb' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
242 bg_to_png ( $fai_file, $type_dir.$type_prefix.'genome_plus.bedgraph', $type_dir.$type_prefix.'genome_minus.bedgraph', $Gviz_genome_rand, 'Mb' );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
243
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
244 #HTML Details
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
245 my $prefix_details_pages = $dir.$fastq_n[$child].'-'.$types_names[$grand_child];
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
246 details_pages ( $type_dir, $prefix_details_pages, \@fastq_n, $fastq_n[$child], $misTE, $dir, $Pcheck );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
247
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
248 $pm2->finish();
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
249 }
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
250 $pm2->wait_all_children;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
251
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
252 if ( $Pcheck eq 'true' )
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
253 {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
254 my $ppp = $group_dir.'PPPartners/'; mkdir $ppp;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
255 print $report "ping_pong_partners $group_dir/piRNAs/TEs.sam $ppp\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
256 ping_pong_partners ( $group_dir.'TEs.fai', $group_dir.'piRNAs/piRNAs-TEs_sorted.bam', $ppp, $pi_min );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
257 my $ppp_page = $dir.$fastq_n[$child].'-piRNAs-PPP.html';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
258 ppp_page ( $group_dir, $ppp_page, \@fastq_n, $fastq_n[$child], $ppp, $dir );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
259 }
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
260
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
261 #HTML Main Webpage
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
262 my $index_page = $dir.$fastq_n[$child].'.html';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
263 main_page ( $gen_dir, $index_page, \@fastq_n, $fastq_n[$child], $ma, $ma_uni, $dir );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
264 copy ($index_page, $html_out) if $child == 0;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
265 #HTML Menu
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
266 my $menu_page = $dir.$fastq_n[$child].'-sub.html';
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
267 menu_page ( $group_dir, $menu_page, \@fastq_n, $fastq_n[$child], $min, $max, $si_min, $si_max, $pi_min, $pi_max, $dir );
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
268 unlink glob "'$group_dir'*.sam"; unlink glob "'$group_dir'*.fastq";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
269 $pm->finish(); # pass an exit code to finish
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
270 }
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
271 $pm->wait_all_children;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
272 unlink glob "'$dir'"."dataset_*symlink.fa*";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
273 print $report "Job done!\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
274 close $report;
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
275 } else {
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
276 print "sRNAPipe v1.1\n";
9185ca0a7b43 Updated package according to recommendations.
pierre.pouchin
parents: 60
diff changeset
277 }