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