annotate bin/TXdbgen.pl @ 5:2ebca9da5e42 draft default tip

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