|
16
|
1 #!/usr/bin/perl -w
|
|
|
2 #Filename:
|
|
|
3 #Author: chentt
|
|
|
4 #Email:
|
|
|
5 #Date: 2012-4-6
|
|
|
6 #Modified:
|
|
|
7 #Description:
|
|
|
8 my $version=1.00;
|
|
|
9
|
|
|
10 use strict;
|
|
|
11 use Getopt::Long;
|
|
|
12
|
|
|
13 my %opts;
|
|
|
14 GetOptions(\%opts,"i=s","o=s","h");
|
|
|
15 if (!(defined $opts{i} and defined $opts{o}) || defined $opts{h}) { #necessary arguments
|
|
|
16 &usage;
|
|
|
17 }
|
|
|
18 open IN,"<$opts{i}";
|
|
|
19 open OUT ,">$opts{o}";
|
|
|
20 print OUT "#ID\tchr\tstart\tend\tstrand\n";
|
|
|
21 my $n=1;
|
|
|
22 my %gene1;
|
|
|
23 while (my $aline=<IN>) {
|
|
|
24 chomp $aline;
|
|
|
25 next if($aline=~/^\#/);
|
|
|
26 my @tmp=split/\t/,$aline;
|
|
|
27 my $ID;
|
|
|
28 if ($tmp[2] eq "gene") {
|
|
|
29 $tmp[0]=~s/Chr\./Chr/;
|
|
|
30 #$tmp[0]=~s/Chr/chr/;
|
|
|
31 my @infor=split/;/,$tmp[8];
|
|
|
32 for (my $i=0;$i<@infor ;$i++) {
|
|
|
33 if ($infor[$i]=~/Alias\=(\S+)$/) {
|
|
|
34 $ID=$1;
|
|
|
35 last;
|
|
|
36 }
|
|
|
37 else {
|
|
|
38 $ID="unknown$n";
|
|
|
39 $n++;
|
|
|
40 }
|
|
|
41 }
|
|
|
42 #$gene{$tmp[0]}{$ID}=[$tmp[3],$tmp[4],$tmp[6]];#$gene{chr}{geneID}=[start,end,strand]
|
|
|
43 push @{$gene1{$ID}},[$tmp[3],$tmp[4],$tmp[0]];
|
|
|
44 print OUT "$ID\t$tmp[0]\t$tmp[3]\t$tmp[4]\t$tmp[6]\n";
|
|
|
45 }
|
|
|
46 }
|
|
|
47 close IN;
|
|
|
48 close OUT;
|
|
|
49
|
|
|
50
|
|
|
51 sub usage{
|
|
|
52 print <<"USAGE";
|
|
|
53 Version $version
|
|
|
54 Usage:
|
|
|
55 $0 -i -o -h
|
|
|
56 options:
|
|
|
57 -i input cluster file
|
|
|
58 -o output file
|
|
|
59 -h help
|
|
|
60 USAGE
|
|
|
61 exit(1);
|
|
|
62 } |
