bin/jbrowse_vcf_tools/create_indiv.pl

   1 #!/usr/bin/perl
   2
   3 use warnings;
   4 use strict;
   5 use Getopt::Std;
   6
   7 our ($opt_v, $opt_o);
   8
   9 getopts('v:o:');
  10
  11 my $raw_vcf_file = $opt_v;
  12 my $output_dir = $opt_o;
  13 my @header;
  14 my $vcf_file;
  15 my %already_seen;
  16 my $rep;
  17 my $rep_num = '';
  18 my $accession_name;
  19
  20 open (SNPS, "<", $raw_vcf_file) || die "Can't open $raw_vcf_file!\n";
  21
  22 while (<SNPS>) {
  23
  24 #------------------------------------------------------------------
  25 # Save header info that should be present in all files into an array
  26 #-------------------------------------------------------------------
  27
  28     if ($. <= 10) {
  29         print $. . "\n";
  30         push @header, $_;
  31
  32     } elsif (m/^#CHROM/) {
  33         last;
  34     } else {
  35
  36 #--------------------------------------------------------------------------------------------------
  37 # Take line with individual accession name and extract that name. Use name to create output vcf file
  38 #---------------------------------------------------------------------------------------------------
  39
  40         my $full_line = $_;
  41         chomp ($accession_name = $full_line);
  42         print "Working on accession $accession_name\n";
  43         if (exists $already_seen{$accession_name}) {     # check to see if this is a repetition
  44             print STDERR "Rep found for $accession_name !\n";
  45             $rep_num = $already_seen{$accession_name};
  46             $rep_num++;
  47             $already_seen{$accession_name} = $rep_num;
  48             $rep = "_rep" . $rep_num;
  49         } else {
  50             $already_seen{$accession_name} = 1;
  51         }
  52
  53         my $column_number = $.;
  54         $column_number--;
  55
  56         $vcf_file = $output_dir . "/" . $accession_name . "_2015_V6" . $rep . ".vcf_" . $column_number;
  57         $rep='';
  58
  59         print $vcf_file . "\n";
  60         open (OUT, ">", $vcf_file) || die "Can't open $vcf_file\n";
  61
  62 #--------------------------------------------------------------
  63 # Print header and info about a single accession to output file
  64 #--------------------------------------------------------------
  65
  66         foreach my $line (@header) {
  67             print OUT $line;
  68         }
  69     }
  70 }