bin/combine_dosage_files.pl

   1
   2 use strict;
   3
   4 use Data::Dumper;
   5 use Getopt::Std;
   6 use CXGN::GenotypeIO;
   7
   8 our ($opt_r, $opt_o); # -r: fix r headers -o: outfile
   9 getopts('ro:');
  10
  11 my @files = @ARGV;
  12
  13 my @io = ();
  14 open(my $OUT, ">", $opt_o) || die "Can't open outfile $opt_o";
  15
  16 foreach my $f (@files) {
  17     print STDERR "Opening file $f... ";
  18     my $gtio = CXGN::GenotypeIO->new( { file => $f, format=>"dosage_transposed", fix_r_headers => $opt_r });
  19     push @io, $gtio;
  20
  21     print STDERR "Done.\n";
  22 }
  23
  24 my %all_markers;
  25 my %all_accs;
  26
  27 foreach my $io (@io) {
  28     print STDERR "processing file ".$io->plugin->file().".\n";
  29     while (my $gt= $io->next()) {
  30         my $name = $gt->name();
  31         my $markers = $gt->markers();
  32         foreach my $m (@$markers) {
  33             $all_markers{$m}++;
  34         }
  35         my $scores = $gt->rawscores();
  36         #print STDERR Dumper($scores);
  37         foreach my $k (keys %$scores) {
  38             #print "Adding score $scores->{$k} for marker $k to accession $name...\n";
  39             $all_accs{$name}->{$k} = $scores->{$k};
  40         }
  41     }
  42 }
  43
  44
  45 foreach my $m (sort keys %all_markers) {
  46     print $OUT "\t".$m;
  47 }
  48 print $OUT "\n";
  49
  50
  51 foreach my $name (sort keys %all_accs) {
  52     print $OUT $name;
  53     foreach my $m (sort keys %all_markers) {
  54         print $OUT "\t".$all_accs{$name}->{$m};
  55     }
  56     print $OUT "\n";
  57 }
  58
  59
  60 print STDERR "Done.\n";