lib/CXGN/SNPsIO.pm

   1
   2 =head NAME
   3
   4 CXGN::SNPsIO;
   5
   6 =head DESCRIPTION
   7
   8 =head AUTHOR
   9
  10 =cut
  11
  12 package CXGN::SNPsIO;
  13
  14 use Moose;
  15
  16 use Data::Dumper;
  17 use IO::File;
  18 use CXGN::Genotype::SNP;
  19
  20 has 'file' => ( isa => 'Str',
  21                 is  => 'rw',
  22                 required => 1,
  23     );
  24
  25 has 'fh' => (isa => 'FileHandle',
  26              is => 'rw',
  27     );
  28
  29 has 'header' => ( isa => 'Str',
  30                   is  => 'rw',
  31     );
  32
  33 has 'accessions'  => ( isa => 'ArrayRef',
  34                   is  => 'rw',
  35     );
  36
  37 has 'ignore_accessions' => (isa => 'HashRef',
  38                             is => 'rw',
  39     );
  40
  41 has 'valid_accessions' => (isa => 'ArrayRef',
  42                            is => 'rw',
  43     );
  44
  45 has 'filter' => ( isa => 'Str',
  46                   is  => 'rw',
  47     );
  48
  49
  50
  51 sub BUILD {
  52     my $self = shift;
  53     my $args = shift;
  54
  55     my $fh = IO::File->new($args->{file});
  56
  57     while (<$fh>) {
  58         chomp;
  59
  60         if (m/^\#\#/) {
  61             next;
  62         }
  63
  64         if (m/^\#CHROM/) {
  65             chomp($_);
  66             #print STDERR "found header $_\n";
  67             $self->header($_);
  68             # CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT SNPS..
  69             my @fields = split /\s+/, $_;
  70             chomp(@fields);
  71             my @accessions = @fields[9..$#fields];
  72             $self->accessions(\@accessions);
  73             print STDERR "ACCESSIONS: ".join("|", @accessions)."\n";
  74             last;
  75         }
  76     }
  77     $self->fh($fh);
  78 }
  79
  80 sub next_line {
  81     my $self = shift;
  82
  83     my $line = "";
  84     my $fh = $self->fh();
  85     if ($line = <$fh>) {
  86         chomp($line);
  87         return $line;
  88     }
  89     else {
  90         if (! $self->header() ) {
  91             die "Header not seen. This file may have format problems.\n";
  92         }
  93         return undef;
  94     }
  95 }
  96
  97 sub next {
  98     my $self = shift;
  99     my $line = $self->next_line();
 100
 101     if (defined($line)) {
 102
 103         my ($chr, $position, $snp_id, $ref_allele, $alt_allele, $qual, $filter, $info, $format,  @snps) = split /\t/, $line;
 104         #print STDERR "Processing $snp_id\n";
 105         my $snps = CXGN::SNPs->new( );
 106         $snps->raw($line);
 107         $snps->id($snp_id);
 108         $snps->chr($chr);
 109         $snps->position($position);
 110         $snps->ref_allele($ref_allele);
 111         $snps->alt_allele($alt_allele);
 112         $snps->qual($qual);
 113         $snps->filter($filter);
 114         my $depth = $info;
 115         $depth =~ s/DP=(\d+)/$1/i;
 116         if ($depth eq ".") { $depth=0; }
 117         $snps->depth($depth);
 118         $snps->info($info);
 119         $snps->format($format);
 120         $snps->accessions($self->accessions());
 121         $snps->valid_accessions($self->accessions());
 122         #print STDERR "ACCESSIONS: ".(join ",", @{$self->{accessions}})."\n";
 123         my %snp_objects = ();
 124         for(my $n=0; $n<@snps; $n++) {
 125             my $snp = CXGN::Genotype::SNP->new( { id=>$snp_id, format => $format, vcf_string => $snps[$n] } );
 126
 127             my $accession = $self->accessions()->[$n]."\n";
 128             chomp($accession);
 129             #print "ACCESSION: $accession\n";
 130             #if ($accession =~ m/1002:250060174/) { print STDERR "Saw accession 1002:250060174" }
 131             $snp->accession($accession);
 132             $snp_objects{$accession} = $snp;
 133         }
 134         #print STDERR Dumper(\%snp_objects);
 135         $snps->snps(\%snp_objects);
 136         return $snps;
 137     }
 138     else {
 139         #print STDERR "LINE NOT DEFINED.\n";
 140         return undef;
 141     }
 142 }
 143
 144 sub close {
 145     my $self = shift;
 146     close($self->fh());
 147 }
 148
 149 sub total_accessions {
 150     my $self = shift;
 151     return scalar(@{$self->accessions});
 152 }
 153
 154
 155 __PACKAGE__->meta->make_immutable;
 156
 157 1;