bismark: bismark_deduplicate/deduplicate

annotate bismark_deduplicate/deduplicate_bismark @ 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

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 strict;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	3 use warnings;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	4 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	5
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	6
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	7 ### This script is supposed to remove alignments to the same position in the genome which can arise by e.g. PCR amplification
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	8 ### Paired-end alignments are considered a duplicate if both partner reasd start and end at the exact same position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	9
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	10 ### May 13, 2013
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	11 ### Changed the single-end trimming behavior so that only the start coordinate will be used. This avoids duplicate reads that have been trimmed to a varying extent
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	12 ### Changed the way of determining the end of reads in SAM format to using the CIGAR string if the read contains InDels
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	13
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	14 ### 16 July 2013
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	15 ### Adding a new deduplication mode for barcoded RRBS-Seq
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	16
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	17 ### 27 Sept 2013
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	18 ### Added close statement for all output filehandles (which should probably have been there from the start...)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	19
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	20 ### 8 Jan 2015
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	21 ### to detect paired-end command from the @PG line we are no requiring spaces before and after the -1 or -2
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	22
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	23 ### 09 Mar 2015
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	24 ### Removing newline characters also from the read conversion flag in case the tags had been reordered and are now present in the very last column
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 ### 19 08 2015
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	27 ### Hiding the option --representative from view to discourage people from using it (it was nearly always not what they wanted to do anyway). It should still work
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	28 ### for alignments that do not contain any indels
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	29 ### Just for reference, here is the the text:
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	30 ### print "--representative\twill browse through all sequences and print out the sequence with the most representative (as in most frequent) methylation call for any given position. Note that this is very likely the most highly amplified PCR product for a given sequence\n\n";
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	33 my $dedup_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	34
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	35 my $help;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	36 my $representative;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	37 my $single;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	38 my $paired;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	39 my $global_single;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	40 my $global_paired;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	41 my $vanilla;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	42 my $samtools_path;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	43 my $bam;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	44 my $rrbs;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	45 my $version;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	46
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	47 my $command_line = GetOptions ('help' => \$help,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	48 'representative' => \$representative,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	49 's\|single' => \$global_single,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	50 'p\|paired' => \$global_paired,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	51 'vanilla' => \$vanilla,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	52 'samtools_path=s' => \$samtools_path,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	53 'bam' => \$bam,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	54 'barcode' => \$rrbs,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	55 'version' => \$version,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	56 );
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	57
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	58 die "Please respecify command line options\n\n" 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	59
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	60 if ($help){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	61 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	62 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	63 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	64
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	65 if ($version){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	66 print << "VERSION";
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 Bismark Deduplication Module
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	69
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	70 Deduplicator Version: $dedup_version
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	71 Copyright 2010-16 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	72 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	73
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	74
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	75 VERSION
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	76 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	77 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	78
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	79
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	80
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	81 my @filenames = @ARGV;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	82
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	83 unless (@filenames){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	84 print "Please provide one or more Bismark output files for deduplication\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	85 sleep (2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	86 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	87 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	88 }
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	91 ### OPTIONS
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	92 unless ($global_single or $global_paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	93 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	94 die "Please specify either -s (single-end) or -p (paired-end) for deduplication. Reading this information from the \@PG header line only works for SAM/BAM 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	95 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	96 warn "\nNeither -s (single-end) nor -p (paired-end) selected for deduplication. Trying to extract this information for each file separately from the \@PG line of the SAM/BAM file\n";
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	99 if ($global_paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	100 if ($global_single){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	101 die "Please select either -s for single-end files or -p for paired-end files, but not both at the same time!\n\n";
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 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	104
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	105 if ($rrbs){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	106 die "Barcode deduplication only works with Bismark SAM (or BAM) output (in attempt to phase out the vanilla format)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	107 }
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 warn "Processing paired-end custom Bismark output 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	110 warn join ("\t",@filenames),"\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 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	113 warn "Processing paired-end Bismark output file(s) (SAM format):\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 join ("\t",@filenames),"\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	115 }
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 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	118 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	119 warn "Processing single-end custom Bismark output 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	120 warn join ("\t",@filenames),"\n\n";
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 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	123 warn "Processing single-end Bismark output file(s) (SAM format):\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	124 warn join ("\t",@filenames),"\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	125 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	126 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	127
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	128 ### PATH TO SAMTOOLS
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	129 if (defined $samtools_path){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	130 # if Samtools was specified as full command
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	131 if ($samtools_path =~ /samtools$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	132 if (-e $samtools_path){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	133 # Samtools executable found
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	134 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	135 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	136 die "Could not find an installation of Samtools at the location $samtools_path. Please respecify\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	137 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	138 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	139 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	140 unless ($samtools_path =~ /\/$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	141 $samtools_path =~ s/$/\//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	142 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	143 $samtools_path .= 'samtools';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	144 if (-e $samtools_path){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	145 # Samtools executable found
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	146 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	147 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	148 die "Could not find an installation of Samtools at the location $samtools_path. Please respecify\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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	151 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	152 # Check whether Samtools is in the PATH if no path was supplied by the user
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	153 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	154 if (!system "which samtools >/dev/null 2>&1"){ # STDOUT is binned, STDERR is redirected to STDOUT. Returns 0 if Samtools is in the PATH
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	155 $samtools_path = `which samtools`;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	156 chomp $samtools_path;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	157 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	158 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	159
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	160 if ($bam){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	161 if (defined $samtools_path){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	162 $bam = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	163 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	164 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	165 warn "No Samtools found on your system, writing out a gzipped SAM file instead\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	166 $bam = 2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	167 }
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 $bam = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	171 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	172
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	173
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	174 if ($representative){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	175 warn "\nIf there are several alignments to a single position in the genome the alignment with the most representative methylation call will be chosen (this might be the most highly amplified PCR product...)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	176 sleep (1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	177 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	178 elsif($rrbs){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	179 warn "\nIf the input is a multiplexed sample with several alignments to a single position in the genome, only alignments with a unique barcode will be chosen)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	180 sleep (1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	181 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	182 else{ # default; random (=first) alignment
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	183 warn "\nIf there are several alignments to a single position in the genome the first alignment will be chosen. Since the input files are not in any way sorted this is a near-enough random selection of reads.\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	184 sleep (1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	185 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	186
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	187 foreach my $file (@filenames){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	188
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	189 if ($global_single){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	190 $paired = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	191 $single = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	192 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	193 elsif($global_paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	194 $paired = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	195 $single = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	196 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	197
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	198 unless($global_single or $global_paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	199
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	200 warn "Trying to determine the type of mapping from the SAM header line\n"; sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	201
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	202 ### if the user did not specify whether the alignment file was single-end or paired-end we are trying to get this information from the @PG header line in the SAM/BAM file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	203 if ($file =~ /\.gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	204 open (DETERMINE,"gunzip -c $file \|") or die "Unable to read from gzipped file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	205 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	206 elsif ($file =~ /\.bam$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	207 open (DETERMINE,"$samtools_path view -h $file \|") or die "Unable to read from BAM file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	208 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	209 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	210 open (DETERMINE,$file) or die "Unable to read from $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	211 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	212 while (<DETERMINE>){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	213 last unless (/^\@/);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	214 if ($_ =~ /^\@PG/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	215 # warn "found the \@PG line:\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	216 # warn "$_";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	217
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	218 if ($_ =~ /\s+-1\s+/ and $_ =~ /\s+-2\s+/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	219 warn "Treating file as paired-end data (extracted from \@PG line)\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	220 $paired = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	221 $single = 0;
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 warn "Treating file as single-end data (extracted from \@PG line)\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	225 $paired = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	226 $single = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	227 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	228 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	229 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	230 close DETERMINE or warn "$!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	231 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	232
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	233 if ($file =~ /(\.bam$\|\.sam$)/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	234 bam_isEmpty($file);
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
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 ### OPTIONS
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	239 unless ($single or $paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	240 die "Please specify either -s (single-end) or -p (paired-end) for deduplication, or provide a SAM/BAM file that contains the \@PG header line\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	241 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	242
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	243 ###
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	244 unless ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	245 if ($paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	246 test_positional_sorting($file);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	247 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	248 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	249
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	250
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	251 ### writing to a report file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	252 my $report = $file;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	253
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	254 $report =~ s/\.gz$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	255 $report =~ s/\.sam$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	256 $report =~ s/\.bam$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	257 $report =~ s/\.txt$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	258 $report =~ s/$/.deduplication_report.txt/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	259
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	260 open (REPORT,'>',$report) or die "Failed to write to report file to $report: $!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	261
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	262
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	263 ### for representative methylation calls we need to discriminate between single-end and paired-end files as the latter have 2 methylation call strings
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	264 if($representative){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	265 deduplicate_representative($file);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	266 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	267
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	268 elsif($rrbs){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	269 deduplicate_barcoded_rrbs($file);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	270 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	271
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	272 ### as the default option we simply write out the first read for a position and discard all others. This is the fastest option
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	273 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	274
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	275 my %unique_seqs;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	276 my %positions;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	277
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	278 if ($file =~ /\.gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	279 open (IN,"gunzip -c $file \|") or die "Unable to read from gzipped file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	280 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	281 elsif ($file =~ /\.bam$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	282 open (IN,"$samtools_path view -h $file \|") or die "Unable to read from BAM file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	283 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	284 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	285 open (IN,$file) or die "Unable to read from $file: $!\n";
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 my $outfile = $file;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	289 $outfile =~ s/\.gz$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	290 $outfile =~ s/\.sam$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	291 $outfile =~ s/\.bam$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	292 $outfile =~ s/\.txt$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	293
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	294 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	295 $outfile =~ s/$/_deduplicated.txt/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	296 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	297 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	298 if ($bam == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	299 $outfile =~ s/$/.deduplicated.bam/;
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 elsif ($bam == 2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	302 $outfile =~ s/$/.deduplicated.sam.gz/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	303 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	304 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	305 $outfile =~ s/$/.deduplicated.sam/;
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 ($bam == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	309 open (OUT,"\| $samtools_path view -bSh 2>/dev/null - > $outfile") or die "Failed to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	310 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	311 elsif($bam == 2){ ### no Samtools found on system. Using GZIP compression instead
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	312 open (OUT,"\| gzip -c - > $outfile") or die "Failed to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	313 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	314
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	315 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	316 open (OUT,'>',$outfile) or die "Unable to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	317 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	318
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	319 ### need to proceed slightly differently for the custom Bismark and Bismark SAM output
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	320 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	321 $_ = <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	322 print OUT; # Printing the Bismark version to the de-duplicated file again
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	323 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	324 my $count = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	325 my $unique_seqs = 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 $removed = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	327
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	328 while (<IN>){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	329
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	330 if ($count == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	331 if ($_ =~ /^Bismark version:/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	332 warn "The file appears to be in the custom Bismark and not SAM format. Please see option --vanilla!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	333 sleep (2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	334 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	335 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	336 }
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	339 ### if this was a SAM file we ignore header lines
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	340 unless ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	341 if (/^\@\w{2}\t/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	342 print "skipping header line:\t$_";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	343 print OUT "$_"; # Printing the header lines again into the de-duplicated file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	344 next;
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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	347
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	348 ++$count;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	349 my $composite; # storing positional data. For single end data we are only using the start coordinate since the end might have been trimmed to different lengths
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	350
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	351 my ($strand,$chr,$start,$end,$cigar);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	352 my $line1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	353
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	354 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	355 ($strand,$chr,$start,$end) = (split (/\t/))[1,2,3,4];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	356 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	357 else{ # SAM format
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	358 ($strand,$chr,$start,$cigar) = (split (/\t/))[1,2,3,5]; # we are assigning the FLAG value to $strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	359
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	360 ### SAM single-end
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	361 if ($single){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	362
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	363 if ($strand == 0 ){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	364 ### read aligned to the forward strand. No action needed
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	365 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	366 elsif ($strand == 16){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	367 ### read is on reverse strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	368
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	369 $start -= 1; # only need to adjust this once
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	370
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	371 # for InDel free matches we can simply use the M number in the CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	372 if ($cigar =~ /^(\d+)M$/){ # linear match
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	373 $start += $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	374 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	375
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	376 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	377 # parsing CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	378 my @len = split (/\D+/,$cigar); # storing the length per operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	379 my @ops = split (/\d+/,$cigar); # storing the operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	380 shift @ops; # remove the empty first element
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	381 die "CIGAR string contained a non-matching number of lengths and operations\n" unless (scalar @len == scalar @ops);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	382
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	383 # warn "CIGAR string; $cigar\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	384 ### determining end position of a read
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	385 foreach my $index(0..$#len){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	386 if ($ops[$index] eq 'M'){ # standard matching bases
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	387 $start += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	388 # warn "Operation is 'M', adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	389 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	390 elsif($ops[$index] eq 'I'){ # insertions do not affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	391 # warn "Operation is 'I', next\n";
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 elsif($ops[$index] eq 'D'){ # deletions do affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	394 # warn "Operation is 'D',adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	395 $start += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	396 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	397 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	398 die "Found CIGAR operations other than M, I or D: '$ops[$index]'. Not allowed at the moment\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	399 }
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 }
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 $composite = join (":",$strand,$chr,$start);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	404 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	405 elsif($paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	406
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	407 ### storing the current line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	408 $line1 = $_;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	409
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	410 my $read_conversion;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	411 my $genome_conversion;
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 while ( /(XR\|XG):Z:([^\t]+)/g ) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	414 my $tag = $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	415 my $value = $2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	416
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	417 if ($tag eq "XR") {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	418 $read_conversion = $value;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	419 $read_conversion =~ s/\r//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	420 chomp $read_conversion;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	421 } elsif ($tag eq "XG") {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	422 $genome_conversion = $value;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	423 $genome_conversion =~ s/\r//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	424 chomp $genome_conversion;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	425 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	426 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	427 die "Failed to determine read and genome conversion from line: $line1\n\n" unless ($read_conversion and $read_conversion);
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 my $index;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	431 if ($read_conversion eq 'CT' and $genome_conversion eq 'CT') { ## original top strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	432 $index = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	433 $strand = '+';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	434 } elsif ($read_conversion eq 'GA' and $genome_conversion eq 'CT') { ## complementary to original top strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	435 $index = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	436 $strand = '-';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	437 } elsif ($read_conversion eq 'GA' and $genome_conversion eq 'GA') { ## complementary to original bottom strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	438 $index = 2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	439 $strand = '+';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	440 } elsif ($read_conversion eq 'CT' and $genome_conversion eq 'GA') { ## original bottom strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	441 $index = 3;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	442 $strand = '-';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	443 } else {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	444 die "Unexpected combination of read and genome conversion: '$read_conversion' / '$genome_conversion'\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	445 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	446
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	447 # if the read aligns in forward orientation we can certainly use the start position of read 1, and only need to work out the end position of read 2
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	448 if ($index == 0 or $index == 2){
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 ### reading in the next line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	451 $_ = <IN>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	452 # the only thing we need is the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	453 ($end,my $cigar_2) = (split (/\t/))[3,5];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	454
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	455 $end -= 1; # only need to adjust this once
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	456
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	457 # for InDel free matches we can simply use the M number in the CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	458 if ($cigar_2 =~ /^(\d+)M$/){ # linear match
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	459 $end += $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	460 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	461 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	462 # parsing CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	463 my @len = split (/\D+/,$cigar_2); # storing the length per operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	464 my @ops = split (/\d+/,$cigar_2); # storing the operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	465 shift @ops; # remove the empty first element
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	466 die "CIGAR string contained a non-matching number of lengths and operations ($cigar_2)\n" unless (scalar @len == scalar @ops);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	467
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	468 # warn "CIGAR string; $cigar_2\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	469 ### determining end position of the read
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	470 foreach my $index(0..$#len){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	471 if ($ops[$index] eq 'M'){ # standard matching bases
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	472 $end += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	473 # warn "Operation is 'M', adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	474 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	475 elsif($ops[$index] eq 'I'){ # insertions do not affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	476 # warn "Operation is 'I', next\n";
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 elsif($ops[$index] eq 'D'){ # deletions do affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	479 # warn "Operation is 'D',adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	480 $end += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	481 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	482 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	483 die "Found CIGAR operations other than M, I or D: '$ops[$index]'. Not allowed at the moment\n";
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 }
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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	488 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	489 # else read 1 aligns in reverse orientation and we need to work out the end of the fragment first, and use the start of the next line
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 $end = $start - 1; # need to adjust this only once
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	492
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	493 # for InDel free matches we can simply use the M number in the CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	494 if ($cigar =~ /^(\d+)M$/){ # linear match
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	495 $end += $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	496 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	497 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	498 # parsing CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	499 my @len = split (/\D+/,$cigar); # storing the length per operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	500 my @ops = split (/\d+/,$cigar); # storing the operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	501 shift @ops; # remove the empty first element
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	502 die "CIGAR string contained a non-matching number of lengths and operations ($cigar)\n" unless (scalar @len == scalar @ops);
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 # warn "CIGAR string; $cigar\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	505 ### determining end position of the read
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	506 foreach my $index(0..$#len){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	507 if ($ops[$index] eq 'M'){ # standard matching bases
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	508 $end += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	509 # warn "Operation is 'M', adding $len[$index] bp\n";
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 elsif($ops[$index] eq 'I'){ # insertions do not affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	512 # warn "Operation is 'I', next\n";
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 elsif($ops[$index] eq 'D'){ # deletions do affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	515 # warn "Operation is 'D',adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	516 $end += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	517 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	518 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	519 die "Found CIGAR operations other than M, I or D: '$ops[$index]'. Not allowed at the moment\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	520 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	521 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	522 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	523
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	524 ### reading in the next line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	525 $_ = <IN>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	526 # the only thing we need is the start position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	527 ($start) = (split (/\t/))[3];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	528 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	529 $composite = join (":",$strand,$chr,$start,$end);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	530 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	531
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	532 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	533 die "Input must be single or paired-end\n";
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	537 if (exists $unique_seqs{$composite}){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	538 ++$removed;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	539 unless (exists $positions{$composite}){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	540 $positions{$composite}++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	541 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	542 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	543 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	544 if ($paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	545 unless ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	546 print OUT "$line1"; # printing first paired-end line for SAM output
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	547 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	548 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	549 print OUT "$_"; # printing single-end SAM alignment or second paired-end line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	550 $unique_seqs{$composite}++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	551 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	552 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	553
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	554 my $percentage;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	555 my $percentage_leftover;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	556 my $leftover = $count - $removed;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	557
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	558 unless ($count == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	559 $percentage = sprintf("%.2f",$removed/$count*100);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	560 $percentage_leftover = sprintf("%.2f",$leftover/$count*100);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	561 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	562 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	563 $percentage = 'N/A';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	564 $percentage_leftover = 'N/A';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	565 }
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 warn "\nTotal number of alignments analysed in $file:\t$count\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	568 warn "Total number duplicated alignments removed:\t$removed ($percentage%)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	569 warn "Duplicated alignments were found at:\t",scalar keys %positions," different position(s)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	570 warn "Total count of deduplicated leftover sequences: $leftover ($percentage_leftover% of total)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	571
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	572 print REPORT "\nTotal number of alignments analysed in $file:\t$count\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	573 print REPORT "Total number duplicated alignments removed:\t$removed ($percentage%)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	574 print REPORT "Duplicated alignments were found at:\t",scalar keys %positions," different position(s)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	575 print REPORT "Total count of deduplicated leftover sequences: $leftover ($percentage_leftover% of total)\n\n";
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 close OUT or warn "Failed to close output filehandle: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	579 close REPORT or warn "Failed to close report filehandle: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	580
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	581 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	582
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	583
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	584 sub deduplicate_barcoded_rrbs{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	585
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	586 my $file = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	587
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	588 my %unique_seqs;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	589 my %positions;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	590
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	591 if ($file =~ /\.gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	592 open (IN,"gunzip -c $file \|") or die "Unable to read from gzipped file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	593 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	594 elsif ($file =~ /\.bam$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	595 open (IN,"$samtools_path view -h $file \|") or die "Unable to read from BAM file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	596 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	597 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	598 open (IN,$file) or die "Unable to read from $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	599 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	600
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	601 my $outfile = $file;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	602 $outfile =~ s/\.gz$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	603 $outfile =~ s/\.sam$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	604 $outfile =~ s/\.bam$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	605 $outfile =~ s/\.txt$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	606
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	607 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	608 $outfile =~ s/$/_dedup_RRBS.txt/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	609 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	610 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	611 if ($bam == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	612 $outfile =~ s/$/.dedup_RRBS.bam/;
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 elsif ($bam == 2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	615 $outfile =~ s/$/.dedupRRBS.sam.gz/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	616 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	617 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	618 $outfile =~ s/$/.dedup_RRBS.sam/;
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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	621 if ($bam == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	622 open (OUT,"\| $samtools_path view -bSh 2>/dev/null - > $outfile") or die "Failed to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	623 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	624 elsif($bam == 2){ ### no Samtools found on system. Using GZIP compression instead
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	625 open (OUT,"\| gzip -c - > $outfile") or die "Failed to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	626 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	627 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	628 open (OUT,'>',$outfile) or die "Unable to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	629 }
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 ### This mode only supports Bismark SAM output
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	632 my $count = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	633 my $unique_seqs = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	634 my $removed = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	635
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	636 while (<IN>){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	637
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	638 if ($count == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	639 if ($_ =~ /^Bismark version:/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	640 warn "The file appears to be in the custom Bismark and not SAM format. Please see option --vanilla!\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 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	643 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	644 }
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 ### if this was a SAM file we ignore header lines
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	648 if (/^\@\w{2}\t/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	649 warn "skipping SAM header line:\t$_";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	650 print OUT; # Printing the header lines again into the de-duplicated file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	651 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	652 }
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 ++$count;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	655 my $composite; # storing positional data. For single end data we are only using the start coordinate since the end might have been trimmed to different lengths
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	656 ### in this barcoded mode we also store the read barcode as additional means of assisting the deduplication
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	657 ### in effect the $composite string looks like this (separated by ':'):
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	658
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	659 ### FLAG:chromosome:start:barcode
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	660
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	661 my $end;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	662 my $line1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	663
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	664 # SAM format
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	665 my ($id,$strand,$chr,$start,$cigar) = (split (/\t/))[0,1,2,3,5]; # we are assigning the FLAG value to $strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	666
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	667 $id =~ /:(\w+)$/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	668 my $barcode = $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	669 unless ($barcode){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	670 die "Failed to extract a barcode from the read ID (last element of each read ID needs to be the barcode sequence, e.g. ':CATG'\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	671 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	672
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	673 ### SAM single-end
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	674 if ($single){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	675
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	676 if ($strand == 0 ){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	677 ### read aligned to the forward strand. No action needed
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	678 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	679 elsif ($strand == 16){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	680 ### read is on reverse strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	681
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	682 $start -= 1; # only need to adjust this once
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	683
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	684 # for InDel free matches we can simply use the M number in the CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	685 if ($cigar =~ /^(\d+)M$/){ # linear match
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	686 $start += $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	687 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	688 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	689 # parsing CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	690 my @len = split (/\D+/,$cigar); # storing the length per operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	691 my @ops = split (/\d+/,$cigar); # storing the operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	692 shift @ops; # remove the empty first element
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	693 die "CIGAR string contained a non-matching number of lengths and operations\n" unless (scalar @len == scalar @ops);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	694
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	695 # warn "CIGAR string; $cigar\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	696 ### determining end position of a read
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	697 foreach my $index(0..$#len){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	698 if ($ops[$index] eq 'M'){ # standard matching bases
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	699 $start += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	700 # warn "Operation is 'M', adding $len[$index] bp\n";
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 elsif($ops[$index] eq 'I'){ # insertions do not affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	703 # warn "Operation is 'I', next\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	704 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	705 elsif($ops[$index] eq 'D'){ # deletions do affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	706 # warn "Operation is 'D',adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	707 $start += $len[$index];
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 die "Found CIGAR operations other than M, I or D: '$ops[$index]'. Not allowed at the moment\n";
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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	714 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	715
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	716 ### Here we take the barcode sequence into consideration
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	717 $composite = join (":",$strand,$chr,$start,$barcode);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	718 # warn "$composite\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	719 # sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	720 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	721 elsif($paired){
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 ### storing the current line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	724 $line1 = $_;
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 my $read_conversion;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	727 my $genome_conversion;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	728
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	729 while ( /(XR\|XG):Z:([^\t]+)/g ) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	730 my $tag = $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	731 my $value = $2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	732
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	733 if ($tag eq "XR") {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	734 $read_conversion = $value;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	735 $read_conversion =~ s/\r//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	736 chomp $read_conversion;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	737 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	738 elsif ($tag eq "XG") {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	739 $genome_conversion = $value;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	740 $genome_conversion =~ s/\r//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	741 chomp $genome_conversion;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	742 }
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 die "Failed to determine read and genome conversion from line: $line1\n\n" unless ($read_conversion and $read_conversion);
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	747 my $index;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	748 if ($read_conversion eq 'CT' and $genome_conversion eq 'CT') { ## original top strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	749 $index = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	750 $strand = '+';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	751 } elsif ($read_conversion eq 'GA' and $genome_conversion eq 'CT') { ## complementary to original top strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	752 $index = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	753 $strand = '-';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	754 } elsif ($read_conversion eq 'GA' and $genome_conversion eq 'GA') { ## complementary to original bottom strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	755 $index = 2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	756 $strand = '+';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	757 } elsif ($read_conversion eq 'CT' and $genome_conversion eq 'GA') { ## original bottom strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	758 $index = 3;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	759 $strand = '-';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	760 } else {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	761 die "Unexpected combination of read and genome conversion: '$read_conversion' / '$genome_conversion'\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	762 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	763
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	764 # if the read aligns in forward orientation we can certainly use the start position of read 1, and only need to work out the end position of read 2
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	765 if ($index == 0 or $index == 2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	766
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	767 ### reading in the next line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	768 $_ = <IN>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	769 # the only thing we need is the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	770 ($end,my $cigar_2) = (split (/\t/))[3,5];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	771
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	772 $end -= 1; # only need to adjust this once
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	773
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	774 # for InDel free matches we can simply use the M number in the CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	775 if ($cigar_2 =~ /^(\d+)M$/){ # linear match
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	776 $end += $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	777 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	778 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	779 # parsing CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	780 my @len = split (/\D+/,$cigar_2); # storing the length per operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	781 my @ops = split (/\d+/,$cigar_2); # storing the operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	782 shift @ops; # remove the empty first element
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	783 die "CIGAR string contained a non-matching number of lengths and operations ($cigar_2)\n" unless (scalar @len == scalar @ops);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	784
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	785 # warn "CIGAR string; $cigar_2\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	786 ### determining end position of the read
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	787 foreach my $index(0..$#len){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	788 if ($ops[$index] eq 'M'){ # standard matching bases
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	789 $end += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	790 # warn "Operation is 'M', adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	791 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	792 elsif($ops[$index] eq 'I'){ # insertions do not affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	793 # warn "Operation is 'I', next\n";
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 elsif($ops[$index] eq 'D'){ # deletions do affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	796 # warn "Operation is 'D',adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	797 $end += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	798 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	799 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	800 die "Found CIGAR operations other than M, I or D: '$ops[$index]'. Not allowed at the moment\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	801 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	802 }
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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	805 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	806 # else read 1 aligns in reverse orientation and we need to work out the end of the fragment first, and use the start of the next line
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 $end = $start - 1; # need to adjust this only once
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	809
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	810 # for InDel free matches we can simply use the M number in the CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	811 if ($cigar =~ /^(\d+)M$/){ # linear match
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	812 $end += $1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	813 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	814 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	815 # parsing CIGAR string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	816 my @len = split (/\D+/,$cigar); # storing the length per operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	817 my @ops = split (/\d+/,$cigar); # storing the operation
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	818 shift @ops; # remove the empty first element
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	819 die "CIGAR string contained a non-matching number of lengths and operations ($cigar)\n" unless (scalar @len == scalar @ops);
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 # warn "CIGAR string; $cigar\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	822 ### determining end position of the read
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	823 foreach my $index(0..$#len){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	824 if ($ops[$index] eq 'M'){ # standard matching bases
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	825 $end += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	826 # warn "Operation is 'M', adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	827 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	828 elsif($ops[$index] eq 'I'){ # insertions do not affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	829 # warn "Operation is 'I', next\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	830 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	831 elsif($ops[$index] eq 'D'){ # deletions do affect the end position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	832 # warn "Operation is 'D',adding $len[$index] bp\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	833 $end += $len[$index];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	834 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	835 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	836 die "Found CIGAR operations other than M, I or D: '$ops[$index]'. Not allowed at the moment\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	837 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	838 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	839 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	840
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	841 ### reading in the next line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	842 $_ = <IN>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	843 # the only thing we need is the start position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	844 ($start) = (split (/\t/))[3];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	845 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	846
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	847 ### Here we take the barcode sequence into consideration
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	848 $composite = join (":",$strand,$chr,$start,$end,$barcode);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	849 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	850 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	851 die "Input must be single or paired-end\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	852 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	853
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	854 if (exists $unique_seqs{$composite}){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	855 ++$removed;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	856 unless (exists $positions{$composite}){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	857 $positions{$composite}++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	858 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	859 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	860 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	861 if ($paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	862 print OUT $line1; # printing first paired-end line for SAM output
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	863 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	864 print OUT; # printing single-end SAM alignment or second paired-end line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	865 $unique_seqs{$composite}++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	866 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	867 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	868
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	869 my $percentage;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	870 my $percentage_leftover;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	871 my $leftover = $count - $removed;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	872
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	873 unless ($count == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	874 $percentage = sprintf("%.2f",$removed/$count*100);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	875 $percentage_leftover = sprintf("%.2f",$leftover/$count*100);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	876 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	877 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	878 $percentage = 'N/A';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	879 $percentage_leftover = 'N/A';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	880 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	881
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	882
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	883 warn "\nTotal number of alignments analysed in $file:\t$count\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	884 warn "Total number duplicated alignments removed:\t$removed ($percentage%)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	885 warn "Duplicated alignments were found at:\t",scalar keys %positions," different position(s)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	886 warn "Total count of deduplicated leftover sequences: $leftover ($percentage_leftover% of total)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	887
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	888 print REPORT "\nTotal number of alignments analysed in $file:\t$count\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	889 print REPORT "Total number duplicated alignments removed:\t$removed ($percentage%)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	890 print REPORT "Duplicated alignments were found at:\t",scalar keys %positions," different position(s)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	891 print REPORT "Total count of deduplicated leftover sequences: $leftover ($percentage_leftover% of total)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	892
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	893 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	894
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	895 sub bam_isEmpty{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	896
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	897 my $file = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	898
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	899 if ($file =~ /\.bam$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	900 open (EMPTY,"$samtools_path view $file \|") or die "Unable to read from BAM file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	901 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	902 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	903 open (EMPTY,$file) or die "Unable to read from $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	904 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	905 my $count = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	906 while (<EMPTY>){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	907 if ($_){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	908 $count++; # file contains data, fine.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	909 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	910 last; # one line is enough
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	911 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	912
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	913 if ($count == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	914 die "\n### File appears to be empty, terminating deduplication process. Please make sure the input file has not been truncated. ###\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	915 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	916 close EMPTY or warn "$!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	917 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	918
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	919
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	920 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	921 print "\n",'='x111,"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	922 print "\nThis script is supposed to remove alignments to the same position in the genome from the Bismark mapping output\n(both single and paired-end SAM files), which can arise by e.g. excessive PCR amplification. If sequences align\nto the same genomic position but on different strands they will be scored individually.\n\nNote that deduplication is not recommended for RRBS-type experiments!\n\nIn the default mode, the first alignment to a given position will be used irrespective of its methylation call\n(this is the fastest option, and as the alignments are not ordered in any way this is also near enough random).\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	923 print "For single-end alignments only use the start coordinate of a read will be used for deduplication.\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	924 print "For paired-end alignments the start-coordinate of the first read and the end coordinate of the second\nread will be used for deduplication. ";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	925 print "This script expects the Bismark output to be in SAM format\n(Bismark v0.6.x or higher). To deduplicate the old custom Bismark output please specify '--vanilla'.\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	926 print "* Please note that for paired-end BAM files the deduplication script expects Read1 and Read2 to\nfollow each other in consecutive lines! If the file has been sorted by position make sure that you resort it\nby read name first (e.g. using samtools sort -n) *\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	927
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	928 print '='x111,"\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	929 print ">>> USAGE: ./deduplicate_bismark_alignment_output.pl [options] filename(s) <<<\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	930
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	931 print "-s/--single\t\tdeduplicate single-end Bismark files (default format: SAM)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	932 print "-p/--paired\t\tdeduplicate paired-end Bismark files (default format: SAM)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	933 print "--vanilla\t\tThe input file is in the old custom Bismark format and not in SAM format\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	934 print "--barcode\t\tIn addition to chromosome, start position and orientation this will also take a potential barcode into\n consideration while deduplicating. The barcode needs to be the last element of the read ID and separated\n by a ':', e.g.: MISEQ:14:000000000-A55D0:1:1101:18024:2858_1:N:0:CTCCT\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	935 print "--bam\t\t\tThe output will be written out in BAM format instead of the default SAM format. This script will\n\t\t\tattempt to use the path to Samtools that was specified with '--samtools_path', or, if it hasn't\n\t\t\tbeen specified, attempt to find Samtools in the PATH. If no installation of Samtools can be found,\n\t\t\tthe SAM output will be compressed with GZIP instead (yielding a .sam.gz output file)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	936 print "--samtools_path\t\tThe path to your Samtools installation, e.g. /home/user/samtools/. Does not need to be specified\n\t\t\texplicitly if Samtools is in the PATH already\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	937 print "--version\t\tPrint version information and exit\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	938
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	939 print '='x111,"\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	940
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	941 print "This script was last modified on November 04, 2015\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	942 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	943
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	944
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	945
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	946
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	947 sub test_positional_sorting{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	948
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	949 my $filename = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	950
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	951 print "\nNow testing Bismark result file $filename for positional sorting (which would be bad...)\t";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	952 sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	953
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	954 if ($filename =~ /\.gz$/) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	955 open (TEST,"gunzip -c $filename \|") or die "Can't open gzipped file $filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	956 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	957 elsif ($filename =~ /bam$/ \|\| isBam($filename) ){ ### this would allow to read BAM files that do not end in *.bam
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	958 if ($samtools_path){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	959 open (TEST,"$samtools_path view -h $filename \|") or die "Can't open BAM file $filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	960 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	961 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	962 die "Sorry couldn't find an installation of Samtools. Either specifiy an alternative path using the option '--samtools_path /your/path/', or use a SAM file instead\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	963 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	964 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	965 else {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	966 open (TEST,$filename) or die "Can't open file $filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	967 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	968
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	969 my $count = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	970
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	971 while (<TEST>) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	972 if (/^\@/) { # testing header lines if they contain the @SO flag (for being sorted)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	973 if (/^\@SO/) {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	974 die "SAM/BAM header line '$_' indicates that the Bismark aligment file has been sorted by chromosomal positions which is is incompatible with correct methylation extraction. Please use an unsorted file instead (e.g. use samtools sort -n)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	975 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	976 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	977 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	978 $count++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	979
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	980 last if ($count > 100000); # else we test the first 100000 sequences if they start with the same read ID
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	981
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	982 my ($id_1) = (split (/\t/));
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	983
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	984 ### reading the next line which should be read 2
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	985 $_ = <TEST>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	986 my ($id_2) = (split (/\t/));
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	987 last unless ($id_2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	988 ++$count;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	989
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	990 if ($id_1 eq $id_2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	991 ### ids are the same
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	992 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	993 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	994 else{ ### in previous versions of Bismark we appended /1 and /2 to the read IDs for easier eyeballing which read is which. These tags need to be removed first
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	995 my $id_1_trunc = $id_1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	996 $id_1_trunc =~ s/\/1$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	997 my $id_2_trunc = $id_2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	998 $id_2_trunc =~ s/\/2$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	999
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1000 unless ($id_1_trunc eq $id_2_trunc){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1001 die "\nThe IDs of Read 1 ($id_1) and Read 2 ($id_2) are not the same. This might be a result of sorting the paired-end SAM/BAM files by chromosomal position which is not compatible with correct methylation extraction. Please use an unsorted file instead (e.g. use samtools sort -n)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1002 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1003 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1004 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1005 # close TEST or die $!; somehow fails on our cluster...
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1006 ### If it hasen't died so far then it seems the file is in the correct Bismark format (read 1 and read 2 of a pair directly following each other)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1007 warn "...passed!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1008 sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1009
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1010 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1011
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1012 sub isBam{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1013
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1014 my $filename = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1015
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1016 # reading the first line of the input file to see if it is a BAM file in disguise (i.e. a BAM file that does not end in *.bam which may be produced by Galaxy)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1017 open (DISGUISE,"gunzip -c $filename \|") or die "Failed to open filehandle DISGUISE for $filename\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1018
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1019 ### when BAM files read through a gunzip -c stream they start with BAM...
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1020 my $bam_in_disguise = <DISGUISE>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1021 # warn "BAM in disguise: $bam_in_disguise\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1022
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1023 if ($bam_in_disguise){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1024 if ($bam_in_disguise =~ /^BAM/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1025 close (DISGUISE) or warn "Had trouble closing filehandle BAM in disguise: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1026 return 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1027 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1028 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1029 close (DISGUISE) or warn "Had trouble closing filehandle BAM in disguise: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1030 return 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1031 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1032 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1033 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1034 close (DISGUISE) or warn "Had trouble closing filehandle BAM in disguise: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1035 return 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1036 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1037 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1038
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1039
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1040
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1041
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1042
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1043 #####################################
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1044
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1045 ### The following subroutine "deduplicate_representative" only works correctly for reads that do not contain any indels (as in only Bowtie1-based alignments).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1046 ### Please refrain from using it unless you want to test something out.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1047
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1048 #####################################
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1049
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1050 sub deduplicate_representative {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1051 my $file = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1052
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1053 my %positions;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1054 my %unique_seqs;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1055
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1056 my $count = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1057 my $unique_seqs = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1058 my $removed = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1059
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1060 ### going through the file first and storing all positions as well as their methylation call strings in a hash
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1061 if ($file =~ /\.gz$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1062 open (IN,"gunzip -c $file \|") or die "Unable to read from gzipped file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1063 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1064 elsif ($file =~ /\.bam$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1065 open (IN,"$samtools_path view -h $file \|") or die "Unable to read from BAM file $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1066 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1067 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1068 open (IN,$file) or die "Unable to read from $file: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1069 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1070
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1071 if ($single){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1072
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1073 my $outfile = $file;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1074 $outfile =~ s/\.gz$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1075 $outfile =~ s/\.sam$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1076 $outfile =~ s/\.bam$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1077 $outfile =~ s/\.txt$//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1078
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1079 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1080 $outfile =~ s/$/.deduplicated_to_representative_sequences.txt/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1081 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1082 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1083 if ($bam == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1084 $outfile =~ s/$/.deduplicated_to_representative_sequences.bam/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1085 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1086 elsif ($bam == 2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1087 $outfile =~ s/$/.deduplicated_to_representative_sequences.sam.gz/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1088 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1089 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1090 $outfile =~ s/$/.deduplicated_to_representative_sequences.sam/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1091 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1092 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1093
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1094 if ($bam == 1){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1095 open (OUT,"\| $samtools_path view -bSh 2>/dev/null - > $outfile") or die "Failed to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1096 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1097 elsif($bam == 2){ ### no Samtools found on system. Using GZIP compression instead
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1098 open (OUT,"\| gzip -c - > $outfile") or die "Failed to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1099 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1100 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1101 open (OUT,'>',$outfile) or die "Unable to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1102 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1103
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1104 warn "Reading and storing all alignment positions\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1105
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1106 ### need to proceed slightly differently for the custom Bismark and Bismark SAM output
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1107 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1108 $_ = <IN>; # Bismark version header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1109 print OUT; # Printing the Bismark version to the de-duplicated file again
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1110 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1111
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1112 while (<IN>){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1113
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1114 if ($count == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1115 if ($_ =~ /^Bismark version:/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1116 warn "The file appears to be in the custom Bismark and not SAM format. Please see option --vanilla!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1117 sleep (2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1118 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1119 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1120 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1121 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1122
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1123 ### if this was a SAM file we ignore header lines
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1124 unless ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1125 if (/^\@\w{2}\t/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1126 warn "skipping SAM header line:\t$_";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1127 print OUT; # Printing the header lines again into the de-duplicated file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1128 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1129 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1130 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1131
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1132 my ($strand,$chr,$start,$end,$meth_call);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1133
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1134 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1135 ($strand,$chr,$start,$end,$meth_call) = (split (/\t/))[1,2,3,4,7];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1136 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1137 else{ # SAM format
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1138
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1139 ($strand,$chr,$start,my $seq,$meth_call) = (split (/\t/))[1,2,3,9,13]; # we are assigning the FLAG value to $strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1140 ### SAM single-end
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1141 $end = $start + length($seq) - 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1142 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1143
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1144 my $composite = join (":",$strand,$chr,$start,$end);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1145
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1146 $count++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1147 $positions{$composite}->{$meth_call}->{count}++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1148 $positions{$composite}->{$meth_call}->{alignment} = $_;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1149 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1150 warn "Stored ",scalar keys %positions," different positions for $count sequences in total (+ and - alignments to the same position are scored individually)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1151 close IN or warn $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1152 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1153
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1154 elsif ($paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1155
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1156 ### we are going to concatenate both methylation call strings from the paired end file to form a joint methylation call string
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1157
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1158 my $outfile = $file;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1159 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1160 $outfile =~ s/$/_deduplicated_to_representative_sequences_pe.txt/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1161 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1162 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1163 $outfile =~ s/$/_deduplicated_to_representative_sequences_pe.sam/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1164 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1165
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1166 open (OUT,'>',$outfile) or die "Unable to write to $outfile: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1167 warn "Reading and storing all alignment positions\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1168
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1169 ### need to proceed slightly differently for the custom Bismark and Bismark SAM output
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1170 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1171 $_ = <IN>; # Bismark version header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1172 print OUT; # Printing the Bismark version to the de-duplicated file again
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1173 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1174
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1175 while (<IN>){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1176
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1177 if ($count == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1178 if ($_ =~ /^Bismark version:/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1179 warn "The file appears to be in the custom Bismark and not SAM format. Please see option --vanilla!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1180 sleep (2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1181 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1182 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1183 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1184 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1185
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1186 ### if this was a SAM file we ignore header lines
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1187 unless ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1188 if (/^\@\w{2}\t/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1189 warn "skipping SAM header line:\t$_";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1190 print OUT; # Printing the header lines again into the de-duplicated file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1191 next;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1192 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1193 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1194
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1195 my ($strand,$chr,$start,$end,$meth_call_1,$meth_call_2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1196 my $line1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1197
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1198 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1199 ($strand,$chr,$start,$end,$meth_call_1,$meth_call_2) = (split (/\t/))[1,2,3,4,7,10];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1200 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1201 else{ # SAM paired-end format
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1202
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1203 ($strand,$chr,$start,$meth_call_1) = (split (/\t/))[1,2,3,13]; # we are assigning the FLAG value to $strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1204
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1205 ### storing the first line (= read 1 alignment)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1206 $line1 = $_;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1207
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1208 ### reading in the next line
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1209 $_ = <IN>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1210 # we only need the end position and the methylation call
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1211 (my $pos,my $seq_2,$meth_call_2) = (split (/\t/))[3,9,13];
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1212 $end = $pos + length($seq_2) - 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1213 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1214
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1215 my $composite = join (":",$strand,$chr,$start,$end);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1216 $count++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1217 my $meth_call = $meth_call_1.$meth_call_2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1218
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1219 $positions{$composite}->{$meth_call}->{count}++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1220 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1221 $positions{$composite}->{$meth_call}->{alignment} = $_;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1222 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1223 else{ # SAM PAIRED-END
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1224 $positions{$composite}->{$meth_call}->{alignment_1} = $line1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1225 $positions{$composite}->{$meth_call}->{alignment_2} = $_;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1226 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1227 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1228 warn "Stored ",scalar keys %positions," different positions for $count sequences in total (+ and - alignments to the same position are scored individually)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1229 close IN or warn $!;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1230 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1231
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1232 ### PRINTING RESULTS
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1233
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1234 ### Now going through all stored positions and printing out the methylation call which is most representative, i.e. the one which occurred most often
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1235 warn "Now printing out alignments with the most representative methylation call(s)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1236
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1237 foreach my $pos (keys %positions){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1238 foreach my $meth_call (sort { $positions{$pos}->{$b}->{count} <=> $positions{$pos}->{$a}->{count} }keys %{$positions{$pos}}){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1239 if ($paired){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1240 if ($vanilla){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1241 print OUT $positions{$pos}->{$meth_call}->{alignment};
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1242 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1243 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1244 print OUT $positions{$pos}->{$meth_call}->{alignment_1}; # SAM read 1
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1245 print OUT $positions{$pos}->{$meth_call}->{alignment_2}; # SAM read 2
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1246 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1247 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1248 else{ # single-end
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1249 print OUT $positions{$pos}->{$meth_call}->{alignment};
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1250 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1251 $unique_seqs++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1252 last; ### exiting once we printed a sequence with the most frequent methylation call for a position
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1253 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1254 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1255
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1256 my $percentage;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1257 unless ($count == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1258 $percentage = sprintf ("%.2f",$unique_seqs*100/$count);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1259 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1260 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1261 $percentage = 'N/A';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1262 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1263
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1264 warn "\nTotal number of alignments analysed in $file:\t$count\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1265 warn "Total number of representative alignments printed from $file in total:\t$unique_seqs ($percentage%)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1266 print REPORT "\nTotal number of alignments analysed in $file:\t$count\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1267 print REPORT "Total number of representative alignments printed from $file in total:\t$unique_seqs ($percentage%)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1268
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1269 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1270
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1271
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	1272

Mercurial > repos > bgruening > bismark

annotate bismark_deduplicate/deduplicate_bismark @ 7:fcadce4d9a06 draft