comparison bin/TXdbgen @ 1:adc0f7765d85 draft

planemo upload
author bioitcore
date Thu, 07 Sep 2017 15:06:58 -0400
parents
children
comparison
equal deleted inserted replaced
0:d4ca551ca300 1:adc0f7765d85
1 #!/usr/bin/perl
2 my $SrcFolder="/home/galaxy/galaxy-dist/tools/SpliceTrap.0.90.1/bin";
3 # this script is to generate TXdb database files from bed/gtf file
4
5 use strict;
6 use Cwd;
7 use Getopt::Long;
8
9 my @programs = ('split','bowtie-build','sort', 'uniq', 'ls','bash','rm','mv','cut','grep','echo');
10 foreach my $program (@programs)
11 {
12 die ("CHECK: $program not found\n") if(system("hash $program >/dev/null"));
13
14 }
15
16
17 my $genomedir = "";
18
19 my $annofilename = "";
20 my $txdbname = "userdefined";
21 my $knownonly = 0;
22 my $gtfinput = 0;
23
24 GetOptions (
25 "g:s"=>\$genomedir,
26 "a:s"=>\$annofilename,
27 "n:s"=>\$txdbname,
28 "gtf"=>\$gtfinput,
29 "knownonly"=>\$knownonly
30 );
31
32 my $InputParaDes=" Usage of the script:
33 -g genome fasta file location
34 -a annotation file (bed/gtf)
35 -n txdb name
36 --gtf specify this if annotation file is in gtf format
37 ";
38
39 if($genomedir eq "" or $annofilename eq "")
40 {
41 print $InputParaDes;
42 exit;
43 }
44
45 $genomedir = Cwd::abs_path($genomedir);
46 $annofilename = Cwd::abs_path($annofilename);
47
48 my $annofilebase = `basename $annofilename`;
49 chomp($annofilebase);
50 #need a cache folder to avoid mess
51
52 my $cachefolder = $annofilebase.".cache";
53
54 if (! -e $cachefolder)
55 {
56 mkdir $cachefolder or die "TXDBGEN: could not create cache folder $cachefolder\n";
57 }
58 if($gtfinput)
59 {
60 print "TXDBGEN: converting gtf file into bed format\n";
61 system ("perl $SrcFolder/gtf2bed.pl $annofilename >$cachefolder/$annofilebase.bed");
62 $annofilename = "$cachefolder/$annofilebase.bed";
63 }
64
65
66 print "TXDBGEN: scan $annofilename for AS events...\n";
67 system("perl $SrcFolder/scanbed2txdb.pl $annofilename $cachefolder/TXdb.tmp");
68 print "TXDBGEN: fetch sequences from $genomedir...\n";
69 system("sort -k1,1 $cachefolder/TXdb.tmp >$cachefolder/TXdb.tmp.sort");
70 #get fasta file list
71 system("ls $genomedir/*.fa >$cachefolder/chr.list");
72
73 system("perl $SrcFolder/get_bed_fa_j.pl $cachefolder/TXdb.tmp.sort $cachefolder/chr.list $cachefolder/out.bed $cachefolder/TXdb.fasta");
74
75 print "TXDBGEN: generate files for parallel computing...\n";
76 if (! -e "$cachefolder/parallel")
77 {
78 mkdir "$cachefolder/parallel" or die "TXDBGEN: could not create $cachefolder/parallel\n";
79 }
80 system("grep L $cachefolder/out.bed >$cachefolder/TXdb.bed");
81 system("rm $cachefolder/out.bed");
82 system("sort $cachefolder/TXdb.tmp.evi >$cachefolder/TXdb.evi");
83 system("rm $cachefolder/TXdb.tmp.evi");
84 system("bash $SrcFolder/splitdb.sh $cachefolder/parallel");
85 print "TXDBGEN: build Bowtie index...\n";
86
87 if (! -e "$cachefolder/btw")
88 {
89 mkdir "$cachefolder/btw" or die "TXDBGEN: could not create $cachefolder/btw\n";
90 }
91 system("bowtie-build $cachefolder/TXdb.fasta $cachefolder/btw/TXdb");
92 system("rm $cachefolder/TXdb.tmp* $cachefolder/chr.list");
93 print "TXDBGEN: Copy files to $SrcFolder/../db/$txdbname\n";
94
95 system("mv $cachefolder $SrcFolder/../db/$txdbname");
96 print "TXDBGEN: Done!\n";
97