|
0
|
1 #This script annotates genomic positions to genes for ingenuity analyses.
|
|
|
2 #Aaron Sarver November 15, 2010
|
|
|
3 #version #1.0
|
|
|
4
|
|
|
5 $in =$ARGV[0];
|
|
|
6 $out = $ARGV[1];
|
|
|
7
|
|
|
8 @lib_list=();
|
|
|
9 %seen=();
|
|
|
10
|
|
|
11 @uniq = ();
|
|
|
12 $libstring = '';
|
|
|
13
|
|
|
14
|
|
|
15
|
|
|
16 #open SOURCE, "< Med-nr-7-0.01.txt";
|
|
|
17 open SOURCE, "< $in";
|
|
|
18 open OUT, "> CIS/input.txt";
|
|
|
19 open FIN, "> $out";
|
|
|
20 while (defined($line = <SOURCE>)) {
|
|
|
21 $count++;
|
|
|
22 chomp $line;
|
|
|
23 @field= split(/\t/, $line);
|
|
|
24 $start = $field[1]-50;
|
|
|
25 $stop = $field[1]+50;
|
|
|
26
|
|
|
27 print OUT "$field[0]\t$start\t$stop\n";
|
|
|
28 }
|
|
|
29 close OUT;
|
|
|
30
|
|
|
31 system('perl lib/feature_finder.pl -i CIS/input.txt -icol 1:2:3:6:13 -f lib/mm9.bed -o CIS/input2.txt');
|
|
|
32
|
|
|
33
|
|
|
34 open SOURCE, "< CIS/input2.txt";
|
|
|
35 while (defined($line = <SOURCE>)) {
|
|
|
36 $count++;
|
|
|
37 chomp $line;
|
|
|
38 @field= split(/\t/, $line);
|
|
|
39 #print "$field[0]\n";
|
|
|
40 @genes= split(/,\s/,$field[3]);
|
|
|
41 foreach $gene (@genes) {
|
|
|
42 #print "$gene\n";
|
|
|
43 push (@lib_list,$gene);
|
|
|
44 }
|
|
|
45 }
|
|
|
46
|
|
|
47
|
|
|
48 foreach $item (@lib_list) {
|
|
|
49 unless ($seen{$item}) {
|
|
|
50 $seen{$item}= 1;
|
|
|
51 push(@uniq,$item)
|
|
|
52 }
|
|
|
53
|
|
|
54 }
|
|
|
55
|
|
|
56 #open OUT, "> ingenuity_nr.txt";
|
|
|
57 foreach $item(@uniq) {
|
|
|
58
|
|
|
59 print FIN "$item\t10\n";
|
|
|
60
|
|
|
61 }
|
|
|
62 close FIN;
|
