annotate NEUMA-1.2.1/calculate_iEUMA.pl @ 0:c44c43d185ef draft default tip

NEUMA-1.2.1 Uploaded
author chawhwa
date Thu, 08 Aug 2013 00:46:13 -0400
parents
children
Ignore whitespace changes - Everywhere: Within whitespace: At end of lines:
rev   line source
0
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
1 #!/usr/bin/perl
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
2 # Compute iEUMA for each isoform from the iUtable file and insertlendis file and print out only isoforms whose iEUMA is larger than 0.
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
3
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
4 if(@ARGV<2) { print "usage: $0 iUtable_file insertlendis_file READLENGTH\n"; exit; }
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
5
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
6 my ($iUtable_file,$insertlendis_file,$READLENGTH) = @ARGV;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
7
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
8 open IUTABLE, $iUtable_file or die "Can't open iUtable file $iUtable_file\n";
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
9 chomp($header = <IUTABLE>);
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
10 @distances = split/\t/,$header;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
11 shift @distances; #'gene'
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
12 while(<IUTABLE>){
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
13 chomp;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
14 my @line = split/\t/;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
15 my $gid = shift @line;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
16 my $transcript = shift @line;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
17 my $gene = "$gid\t$transcript";
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
18 for my $i (0..$#distances){
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
19 $iUtable{$gene}{"$distances[$i]"}=$line[$i];
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
20 }
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
21 }
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
22 close IUTABLE;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
23
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
24
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
25 my $totalfreq=0;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
26 open LENDIS, $insertlendis_file or die "Can't open insertlendis file $insertlendis_file\n";
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
27 while(<LENDIS>){
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
28 chomp;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
29 my ($insert_size, $freq) = split/\t/;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
30 $insertlendis{$insert_size}=$freq;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
31 $totalfreq+=$freq;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
32 }
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
33 close LENDIS;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
34
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
35 print "gene\ttranscript\tiEUMA\n"; #header
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
36 for my $gene (sort keys %iUtable){
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
37 my $EUMA=0;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
38 for my $d (keys %{$iUtable{$gene}}){
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
39 $pUMA = $iUtable{$gene}{$d} * ($insertlendis{$d}/$totalfreq) ;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
40 $EUMA += $pUMA;
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
41 }
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
42 if($EUMA>=0.005) { printf "$gene\t%.2f\n",$EUMA; } # use >=0.005 instead of >0, because I'm plotting two decimals after point. (in bp)
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
43 }
c44c43d185ef NEUMA-1.2.1 Uploaded
chawhwa
parents:
diff changeset
44