bismark: new/bismark_genome_preparation annotate

annotate new/bismark_genome_preparation @ 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/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 Cwd;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	5 # use File::Path qw(rmtree);
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	8
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	9 ## This program is Copyright (C) 2010-15, Felix Krueger (felix.krueger@babraham.ac.uk)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	10
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	11 ## This program is free software: you can redistribute it and/or modify
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	12 ## it under the terms of the GNU General Public License as published by
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	13 ## the Free Software Foundation, either version 3 of the License, or
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	14 ## (at your option) any later version.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	15
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	16 ## This program is distributed in the hope that it will be useful,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	17 ## but WITHOUT ANY WARRANTY; without even the implied warranty of
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	18 ## MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	19 ## GNU General Public License for more details.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	20
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	21 ## You should have received a copy of the GNU General Public License
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	22 ## along with this program. If not, see <http://www.gnu.org/licenses/>.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	23
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	24 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	25 use Cwd;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	26
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	27 my $verbose;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	28 my $help;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	29 my $version;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	30 my $man;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	31 my $path_to_bowtie;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	32 my $multi_fasta;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	33 my $single_fasta;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	34 my $bowtie2;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	35
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	36 my $bismark_version = 'v0.14.3';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	37
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	38 GetOptions ('verbose' => \$verbose,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	39 'help' => \$help,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	40 'man' => \$man,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	41 'version' => \$version,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	42 'path_to_bowtie:s' => \$path_to_bowtie,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	43 'single_fasta' => \$single_fasta,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	44 'bowtie2' => \$bowtie2,
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	45 );
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	46
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	47 if ($help or $man){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	48 print_helpfile();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	49 exit;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	50 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	51
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	52 if ($version){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	53 print << "VERSION";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	54
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	55 Bismark - Bisulfite Mapper and Methylation Caller.
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 Bismark Genome Preparation Version: $bismark_version
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	58 Copyright 2010-15 Felix Krueger, Babraham Bioinformatics
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	59 www.bioinformatics.babraham.ac.uk/projects/
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	60
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	61 VERSION
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 my $genome_folder = shift @ARGV; # mandatory
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	66 my %chromosomes; # checking if chromosome names are unique (required)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	67
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	68 # Ensuring a genome folder has been specified
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	69 if ($genome_folder){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	70 unless ($genome_folder =~ /\/$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	71 $genome_folder =~ s/$/\//;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	72 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	73 $verbose and print "Path to genome folder specified as: $genome_folder\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	74 chdir $genome_folder or die "Could't move to directory $genome_folder. Make sure the directory exists! $!";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	75
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	76 # making the genome folder path abolsolute so it won't break if the path was specified relative
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	77 $genome_folder = getcwd;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	78 unless ($genome_folder =~ /\/$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	79 $genome_folder =~ s/$/\//;
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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	82 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	83 die "Please specify a genome folder to be used for bisulfite conversion\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	84 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	85
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	86
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	87 my $CT_dir;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	88 my $GA_dir;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	89
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	90
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	91 if ($single_fasta){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	92 print "Writing individual genomes out into single-entry fasta files (one per chromosome)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	93 $multi_fasta = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	94 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	95 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	96 print "Writing bisulfite genomes out into a single MFA (multi FastA) 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	97 $single_fasta = 0;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	98 $multi_fasta = 1;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	99 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	100
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	101 my @filenames = create_bisulfite_genome_folders();
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 process_sequence_files ();
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 launch_bowtie_indexer();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	106
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	107 sub launch_bowtie_indexer{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	108 if ($bowtie2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	109 print "Bismark Genome Preparation - Step III: Launching the Bowtie 2 indexer\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	110 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	111 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	112 print "Bismark Genome Preparation - Step III: Launching the Bowtie (1) indexer\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	113 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	114 print "Please be aware that this process can - depending on genome size - take up to several hours!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	115 sleep(5);
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 ### if the path to bowtie was specfified explicitely
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	118 if ($path_to_bowtie){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	119 if ($bowtie2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	120 $path_to_bowtie =~ s/$/bowtie2-build/;
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 $path_to_bowtie =~ s/$/bowtie-build/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	124 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	125 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	126 ### otherwise we assume that bowtie-build 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	127 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	128 if ($bowtie2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	129 $path_to_bowtie = 'bowtie2-build';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	130 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	131 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	132 $path_to_bowtie = 'bowtie-build';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	133 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	134 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	135
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	136 $verbose and print "\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 ### Forking the program to run 2 instances of Bowtie-build or Bowtie2-build (= the Bowtie (1/2) indexer)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	139 my $pid = fork();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	140
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	141 # parent process
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	142 if ($pid){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	143 sleep(1);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	144 chdir $CT_dir or die "Unable to change directory: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	145 $verbose and warn "Preparing indexing of CT converted genome in $CT_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	146 my @fasta_files = <*.fa>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	147 my $file_list = join (',',@fasta_files);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	148 $verbose and print "Parent process: Starting to index C->T converted genome with the following command:\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	149 $verbose and print "$path_to_bowtie -f $file_list BS_CT\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	150
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	151 sleep (11);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	152 exec ("$path_to_bowtie","-f","$file_list","BS_CT");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	153 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	154
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	155 # child process
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	156 elsif ($pid == 0){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	157 sleep(2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	158 chdir $GA_dir or die "Unable to change directory: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	159 $verbose and warn "Preparing indexing of GA converted genome in $GA_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	160 my @fasta_files = <*.fa>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	161 my $file_list = join (',',@fasta_files);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	162 $verbose and print "Child process: Starting to index G->A converted genome with the following command:\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	163 $verbose and print "$path_to_bowtie -f $file_list BS_GA\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	164 $verbose and print "(starting in 10 seconds)\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	165 sleep(10);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	166 exec ("$path_to_bowtie","-f","$file_list","BS_GA");
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 # if the platform doesn't support the fork command we will run the indexing processes one after the other
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	170 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	171 print "Forking process was not successful, therefore performing the indexing sequentially instead\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	172 sleep(10);
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 ### moving to CT genome folder
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	175 $verbose and warn "Preparing to index CT converted genome in $CT_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	176 chdir $CT_dir or die "Unable to change directory: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	177 my @fasta_files = <*.fa>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	178 my $file_list = join (',',@fasta_files);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	179 $verbose and print "$file_list\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(2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	181 system ("$path_to_bowtie","-f","$file_list","BS_CT");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	182 @fasta_files=();
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	183 $file_list= '';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	184
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	185 ### moving to GA genome folder
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	186 $verbose and warn "Preparing to index GA converted genome in $GA_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	187 chdir $GA_dir or die "Unable to change directory: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	188 @fasta_files = <*.fa>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	189 $file_list = join (',',@fasta_files);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	190 $verbose and print "$file_list\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	191 sleep(2);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	192 exec ("$path_to_bowtie","-f","$file_list","BS_GA");
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	193 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	194 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	195
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 sub process_sequence_files {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	198
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	199 my ($total_CT_conversions,$total_GA_conversions) = (0,0);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	200 $verbose and print "Bismark Genome Preparation - Step II: Bisulfite converting reference genome\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	201 sleep (3);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	202
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	203 $verbose and print "conversions performed:\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	204 $verbose and print join("\t",'chromosome','C->T','G->A'),"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	205
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	206
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	207 ### If someone wants to index a genome which consists of thousands of contig and scaffold files we need to write the genome conversions into an MFA file
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	208 ### Otherwise the list of comma separated chromosomes we provide for bowtie-build will get too long for the kernel to handle
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	209 ### This is now the default option
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	210
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	211 if ($multi_fasta){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	212 ### Here we just use one multi FastA file name, append .CT_conversion or .GA_conversion and print all sequence conversions into these files
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	213 my $bisulfite_CT_conversion_filename = "$CT_dir/genome_mfa.CT_conversion.fa";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	214 open (CT_CONVERT,'>',$bisulfite_CT_conversion_filename) or die "Can't write to file $bisulfite_CT_conversion_filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	215
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	216 my $bisulfite_GA_conversion_filename = "$GA_dir/genome_mfa.GA_conversion.fa";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	217 open (GA_CONVERT,'>',$bisulfite_GA_conversion_filename) or die "Can't write to file $bisulfite_GA_conversion_filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	218 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	219
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	220 foreach my $filename(@filenames){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	221 my ($chromosome_CT_conversions,$chromosome_GA_conversions) = (0,0);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	222 open (IN,$filename) or die "Failed to read from sequence 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	223 # warn "Reading chromosome information from $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	224
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	225 ### first line needs to be a fastA header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	226 my $first_line = <IN>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	227 chomp $first_line;
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 ### Extracting chromosome name from the FastA header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	230 my $chromosome_name = extract_chromosome_name($first_line);
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 ### Exiting if a chromosome with the same name was present already
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	233 if (exists $chromosomes{$chromosome_name}){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	234 die "Exiting because chromosome name already exists. Please make sure all chromosomes have a unique name!\n";
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 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	237 $chromosomes{$chromosome_name}++;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	238 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	239
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	240 ### alternatively, chromosomes can be written out into single-entry FastA files. This will only work for genomes with up to a few hundred chromosomes.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	241 unless ($multi_fasta){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	242 my $bisulfite_CT_conversion_filename = "$CT_dir/$chromosome_name";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	243 $bisulfite_CT_conversion_filename =~ s/$/.CT_conversion.fa/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	244 open (CT_CONVERT,'>',$bisulfite_CT_conversion_filename) or die "Can't write to file $bisulfite_CT_conversion_filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	245
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	246 my $bisulfite_GA_conversion_filename = "$GA_dir/$chromosome_name";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	247 $bisulfite_GA_conversion_filename =~ s/$/.GA_conversion.fa/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	248 open (GA_CONVERT,'>',$bisulfite_GA_conversion_filename) or die "Can't write to file $bisulfite_GA_conversion_filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	249 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	250
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	251 print CT_CONVERT ">",$chromosome_name,"_CT_converted\n"; # first entry
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	252 print GA_CONVERT ">",$chromosome_name,"_GA_converted\n"; # first entry
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	253
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	254
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	255 while (<IN>){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	256
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	257 ### in case the line is a new fastA header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	258 if ($_ =~ /^>/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	259 ### printing out the stats for the previous chromosome
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	260 $verbose and print join ("\t",$chromosome_name,$chromosome_CT_conversions,$chromosome_GA_conversions),"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	261 ### resetting the chromosome transliteration counters
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	262 ($chromosome_CT_conversions,$chromosome_GA_conversions) = (0,0);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	263
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	264 ### Extracting chromosome name from the additional FastA header
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	265 $chromosome_name = extract_chromosome_name($_);
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 ### alternatively, chromosomes can be written out into single-entry FastA files. This will only work for genomes with up to a few hundred chromosomes.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	268 unless ($multi_fasta){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	269 my $bisulfite_CT_conversion_filename = "$CT_dir/$chromosome_name";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	270 $bisulfite_CT_conversion_filename =~ s/$/.CT_conversion.fa/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	271 open (CT_CONVERT,'>',$bisulfite_CT_conversion_filename) or die "Can't write to file $bisulfite_CT_conversion_filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	272
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	273 my $bisulfite_GA_conversion_filename = "$GA_dir/$chromosome_name";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	274 $bisulfite_GA_conversion_filename =~ s/$/.GA_conversion.fa/;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	275 open (GA_CONVERT,'>',$bisulfite_GA_conversion_filename) or die "Can't write to file $bisulfite_GA_conversion_filename: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	276 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	277
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	278 print CT_CONVERT ">",$chromosome_name,"_CT_converted\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	279 print GA_CONVERT ">",$chromosome_name,"_GA_converted\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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	282 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	283 my $sequence = uc$_;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	284
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	285 ### (I) First replacing all ambiguous sequence characters (such as M,S,R....) by N (G,A,T,C,N and the line endings \r and \n are added to a character group)
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 $sequence =~ s/[^ATCGN\n\r]/N/g;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	288
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	289 ### (II) Writing the chromosome out into a C->T converted version (equals forward strand conversion)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	290
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	291 my $CT_sequence = $sequence;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	292 my $CT_transliterations_performed = ($CT_sequence =~ tr/C/T/); # converts all Cs into Ts
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	293 $total_CT_conversions += $CT_transliterations_performed;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	294 $chromosome_CT_conversions += $CT_transliterations_performed;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	295
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	296 print CT_CONVERT $CT_sequence;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	297
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	298 ### (III) Writing the chromosome out in a G->A converted version of the forward strand (this is equivalent to reverse-
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	299 ### complementing the forward strand and then C->T converting it)
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 my $GA_sequence = $sequence;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	302 my $GA_transliterations_performed = ($GA_sequence =~ tr/G/A/); # converts all Gs to As on the forward strand
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	303 $total_GA_conversions += $GA_transliterations_performed;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	304 $chromosome_GA_conversions += $GA_transliterations_performed;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	305
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	306 print GA_CONVERT $GA_sequence;
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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	309 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	310 $verbose and print join ("\t",$chromosome_name,$chromosome_CT_conversions,$chromosome_GA_conversions),"\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	311 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	312 close (CT_CONVERT) or die "Failed to close filehandle: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	313 close (GA_CONVERT) or die "Failed to close filehandle: $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	314
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	315
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	316 print "\nTotal number of conversions performed:\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	317 print "C->T:\t$total_CT_conversions\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	318 print "G->A:\t$total_GA_conversions\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	319
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	320 warn "\nStep II - Genome bisulfite conversions - completed\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	321 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	322
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	323 sub extract_chromosome_name {
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	324
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	325 my $header = shift;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	326
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	327 ## Bowtie extracts the first string after the initial > in the FASTA file, so we are doing this as well
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	328
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	329 if ($header =~ s/^>//){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	330 my ($chromosome_name) = split (/\s+/,$header);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	331 return $chromosome_name;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	332 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	333 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	334 die "The specified chromosome file doesn't seem to be in FASTA format as required! $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	335 }
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 sub create_bisulfite_genome_folders{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	339
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	340 $verbose and print "Bismark Genome Preparation - Step I: Preparing folders\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	341
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	342 if ($path_to_bowtie){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	343 unless ($path_to_bowtie =~ /\/$/){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	344 $path_to_bowtie =~ s/$/\//;
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 if (chdir $path_to_bowtie){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	347 if ($bowtie2){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	348 $verbose and print "Path to Bowtie 2 specified: $path_to_bowtie\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	349 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	350 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	351 $verbose and print "Path to Bowtie (1) specified: $path_to_bowtie\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	352 }
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 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	355 die "There was an error with the path to bowtie: $!\n";
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 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	358
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	359 chdir $genome_folder or die "Could't move to directory $genome_folder. Make sure the directory exists! $!";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	360
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	361
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	362 # Exiting unless there are fastA files in the folder
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	363 my @filenames = <*.fa>;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	364
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	365 ### if there aren't any genomic files with the extension .fa we will look for files with the extension .fasta
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	366 unless (@filenames){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	367 @filenames = <*.fasta>;
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	370 unless (@filenames){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	371 die "The specified genome folder $genome_folder does not contain any sequence files in FastA format (with .fa or .fasta file extensions\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	372 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	373
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	374 warn "Bisulfite Genome Indexer version $bismark_version (last modified 19 Sept 2013)\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	375 sleep (3);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	376
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	377 # creating a directory inside the genome folder to store the bisfulfite genomes unless it already exists
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	378 my $bisulfite_dir = "${genome_folder}Bisulfite_Genome/";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	379 unless (-d $bisulfite_dir){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	380 mkdir $bisulfite_dir or die "Unable to create directory $bisulfite_dir $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	381 $verbose and print "Created Bisulfite Genome folder $bisulfite_dir\n";
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 else{
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	384 print "\nA directory called $bisulfite_dir already exists. Bisulfite converted sequences and/or already existing Bowtie (1 or 2) indices will be overwritten!\n\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	385 sleep(5);
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	386 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	387
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	388 chdir $bisulfite_dir or die "Unable to move to $bisulfite_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	389 $CT_dir = "${bisulfite_dir}CT_conversion/";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	390 $GA_dir = "${bisulfite_dir}GA_conversion/";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	391
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	392 # creating 2 subdirectories to store a C->T (forward strand conversion) and a G->A (reverse strand conversion)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	393 # converted version of the genome
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	394 unless (-d $CT_dir){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	395 mkdir $CT_dir or die "Unable to create directory $CT_dir $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	396 $verbose and print "Created Bisulfite Genome folder $CT_dir\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	397 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	398 unless (-d $GA_dir){
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	399 mkdir $GA_dir or die "Unable to create directory $GA_dir $!\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	400 $verbose and print "Created Bisulfite Genome folder $GA_dir\n";
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 # moving back to the original genome folder
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	404 chdir $genome_folder or die "Could't move to directory $genome_folder $!";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	405 # $verbose and print "Moved back to genome folder folder $genome_folder\n";
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	406 warn "\nStep I - Prepare genome folders - completed\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	407 return @filenames;
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	408 }
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	409
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	410 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	411 print << 'HOW_TO';
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	412
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	413
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	414 DESCRIPTION
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	415
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	416 This script is supposed to convert a specified reference genome into two different bisulfite
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	417 converted versions and index them for alignments with Bowtie 1 (default), or Bowtie 2. The first
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	418 bisulfite genome will have all Cs converted to Ts (C->T), and the other one will have all Gs
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	419 converted to As (G->A). Both bisulfite genomes will be stored in subfolders within the reference
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	420 genome folder. Once the bisulfite conversion has been completed the program will fork and launch
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	421 two simultaneous instances of the Bowtie 1 or 2 indexer (bowtie-build or bowtie2-build). Be aware
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	422 that the indexing process can take up to several hours; this will mainly depend on genome size
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	423 and system resources.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	424
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 The following is a brief description of command line options and arguments to control the
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	428 Bismark Genome Preparation:
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
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	431 USAGE: bismark_genome_preparation [options] <arguments>
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	432
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	433
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	434 OPTIONS:
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	435
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	436 --help/--man Displays this help filea and exits.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	437
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	438 --version Displays version information and exits.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	439
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	440 --verbose Print verbose output for more details or debugging.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	441
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	442 --path_to_bowtie </../> The full path to the Bowtie 1 or Bowtie 2 installation on your system
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	443 (depending on which aligner/indexer you intend to use). Unless this path
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	444 is specified it is assumed that Bowtie 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	445
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	446 --bowtie2 This will create bisulfite indexes for Bowtie 2. (Default: Bowtie 1).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	447
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	448 --single_fasta Instruct the Bismark Indexer to write the converted genomes into
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	449 single-entry FastA files instead of making one multi-FastA file (MFA)
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	450 per chromosome. This might be useful if individual bisulfite converted
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	451 chromosomes are needed (e.g. for debugging), however it can cause a
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	452 problem with indexing if the number of chromosomes is vast (this is likely
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	453 to be in the range of several thousand files; the operating system can
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	454 only handle lists up to a certain length, and some newly assembled
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	455 genomes may contain 20000-50000 contigs of scaffold files which do exceed
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	456 this list length limit).
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	457
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	458
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	459 ARGUMENTS:
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 <path_to_genome_folder> The path to the folder containing the genome to be bisulfite converted.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	462 The Bismark Genome Preparation expects one or more fastA files in the folder
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	463 (with the file extension: .fa or .fasta). Specifying this path is mandatory.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	464
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	465
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	466 This script was last modified on 16 Oct 2014.
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	467 HOW_TO
fcadce4d9a06 planemo upload for repository https://github.com/bgruening/galaxytools/tree/master/tools/bismark commit b'e6ee273f75fff61d1e419283fa8088528cf59470\n' bgruening parents: diff changeset	468 }

Mercurial > repos > bgruening > bismark

annotate new/bismark_genome_preparation @ 7:fcadce4d9a06 draft