frmt.pred

#!/usr/bin/perl

use strict;

my $in=shift;

my %g;
open(G, "< data/all.genes");
while(my $i=<G>) { chomp($i); $g{$i}=1; }

open(IN, "< $in");
my %r; my %lr; my $f=1;
while(my $i=<IN>) {
  chomp($i);
  
  $i=~s/Adipose\.vo/Adipose-vo/g;
  $i=~s/Adipose\.sc/Adipose-sc/g;

  my @l=split(/\s+/,$i);

  if ((!($l[1]=~/Intercept/)) && (!($l[1]=~/MAF/))) { 
  #$l[0]=$1;
  my $locus=$l[0]; my $gene;

  if ($l[1]=~/(.*)\.CDS/) { $gene=$1; $f=10; }
  elsif ($l[1]=~/(.*)\.DHS/) { $gene=$1; $f=1; }
  elsif ($l[1]=~/(.*)\.DIST/) { $gene=$1; $f=1; }
  elsif ($l[1]=~/[^\.*]\.(.*$)/) { $gene=$1; $f=1; }

  $gene=~s/\./-/;

  if ($l[2] eq ".") { $l[2]=0; }

  if ($gene && $l[2]>0) { $r{$locus}{$gene}+=$l[2]*$f; }

  #$lr{$locus}+=$l[2]*$f;

}
}

foreach my $loc (keys %r) {
  foreach my $gene (keys %{$r{$loc}}) {
    #my $s = $r{$loc}{$gene}/$lr{$loc};
    if (exists $g{$gene}) { 
    # print "$loc.$tag\t$gene\t$r{$loc}{$gene}\n"; 
    print "$loc\t$gene\t$r{$loc}{$gene}\n";
    }
  }
}