lib/CXGN/Phenotypes/File.pm

   1
   2 =head1 NAME
   3
   4 CXGN::Phenotypes::File - a class to parse out information from the files that phenotype searching returns
   5
   6 =head1 DESCRIPTION
   7
   8 =head1 AUTHOR
   9
  10 Lukas Mueller <lam87@cornell.edu>
  11
  12 =cut
  13
  14 package CXGN::Phenotypes::File;
  15
  16 use Moose;
  17 use Data::Dumper;
  18 use File::Slurp qw | slurp |;
  19
  20 has 'file' => (is => 'rw', isa => 'Str');
  21
  22 has 'factors' => ( is => 'rw', isa => 'ArrayRef' );
  23
  24 has 'traits' => ( is => 'rw', isa => 'ArrayRef' );
  25
  26 has 'levels' => ( is => 'rw', isa => 'HashRef' );
  27
  28 has 'remove_quotes' => (is => 'rw', isa => 'Bool', default => sub { return 1; } );
  29
  30 our $FACTOR_COUNT = 38; # number of columns in the file before traits columns start
  31
  32 sub BUILD {
  33     my $self = shift;
  34     my @lines = slurp($self->file());
  35
  36     my $header = $lines[0];
  37     chomp($header);
  38
  39     my @keys = split("\t", $header);
  40
  41     if ($self->remove_quotes()) {
  42         foreach my $k (@keys) {
  43             #print STDERR "Removing quotes from $k...";
  44             $k=~ s/^\"(.*)\"$/$1/;
  45             #print STDERR "Now $k...\n";
  46         }
  47     }
  48
  49
  50     my @data = ();
  51     my %line = ();
  52     my %levels = ();
  53
  54     for (my $i=1; $i<@lines; $i++) {
  55         my @fields = split /\t/, $lines[$i];
  56         for(my $n=0; $n <@keys; $n++) {
  57             if ($self->remove_quotes()) {
  58                 #print STDERR "Removing quotes from $fields[$n]...";
  59                 $fields[$n]=~ s/^\"(.*)\"$/$1/;
  60                 #print STDERR "Now $fields[$n]...\n";
  61             }
  62
  63             if (exists($fields[$n]) && defined($fields[$n])) {
  64                 $line{$keys[$n]}=$fields[$n];
  65                 if ($n<39) {
  66                     $levels{$keys[$n]}->{fields}->{$fields[$n]}++;
  67                     $levels{$keys[$n]}->{distinct} = scalar(keys(%{$levels{$keys[$n]}->{fields}}));
  68                 }
  69             }
  70         }
  71         push @data, \%line;
  72     }
  73     $self->factors( [ @keys[0..$FACTOR_COUNT] ] );
  74     $self->traits( [ @keys[ $FACTOR_COUNT+1..scalar(@keys) ] ] );
  75
  76     $self->levels(\%levels);
  77
  78 }
  79
  80 sub distinct_levels_for_factor {
  81     my $self = shift;
  82     my $factor = shift;
  83
  84     #print STDERR "LEVELS: ".Dumper($self->levels());
  85     return $self->levels()->{$factor}->{distinct};
  86 }
  87
  88
  89 1;