bin/replace_accessions_in_vcf.pl

   1
   2 use strict;
   3
   4 use Hash::Case::Preserve;
   5
   6 my $vcf_file = shift;
   7 my $equivalence_file = shift;
   8
   9
  10 open(my $E, "<", $equivalence_file) || die "Can't open $equivalence_file\n";
  11
  12 tie my %ids, 'Hash::Case::Preserve';
  13 while (<$E>) {
  14     chomp;
  15
  16     my ($original_id, $modified_id, $new_id, $match_type, $stock_type) = split /\t/;
  17     $ids{$original_id} = $new_id;
  18 }
  19
  20 close($E);
  21
  22 open(my $V, "<", $vcf_file) || die "Can't open $vcf_file\n";
  23
  24 while (<$V>) {
  25     chomp();
  26
  27     if ($_ =~ m/^\#CHROM/) {
  28         print STDERR "Parsing ids in vcf file...\n";
  29         my @F = split /\t/;
  30         my $count = 0;
  31         foreach my $f (@F) {
  32             $count++;
  33             if (exists($ids{$f}) && ($ids{$f} ne $f)) {
  34                 print STDERR "replace $f with $ids{$f} ($count), ";
  35                 $f = $ids{$f};
  36             }
  37         }
  38         my $line = join("\t", @F);
  39         print "$line\n";
  40     }
  41     else {
  42         print $_."\n";
  43     }
  44 }
  45
  46