File Coverage

blib/lib/CracTools/GFF/Annotation.pm

Criterion	Covered	Total	%
statement	88	98	89.8
branch	22	34	64.7
condition	3	6	50.0
subroutine	17	17	100.0
pod	12	12	100.0
total	142	167	85.0

package CracTools::GFF::Annotation;
{
  $CracTools::GFF::Annotation::DIST = 'CracTools';
}
# ABSTRACT: Parse GFF lines.
$CracTools::GFF::Annotation::VERSION = '1.22';
use strict;
use warnings;

use Carp;
use CracTools::Utils;


sub new {
  my $class = shift;
  my $line = shift;
  my $format = shift;
  if(!defined $format) {
    $format = 'gff3';
  }

  my $self = bless {format => $format}, $class;

  if(defined $line) {
    $self->_init($line);
  }

  return $self;
}

sub _init {
  my ($self,$line) = @_;

  # Split the line with TABS
  my ($chr,$source,$feature,$start,$end,$score,$strand,$phase,$attributes) = split("\t",$line);

  # Get the strand if 1/-1 format
  $strand = CracTools::Utils::convertStrand($strand);

  # We do not want any "chr" string before the reference sequence value
  $chr =~ s/^chr//;

  # Loading the 8 first columns
  $self->{chr} = $chr;
  $self->{source} = $source;
  $self->{feature} = $feature;

  # Conversion to 0-based coordinate system
  $self->{start} = $start-1;
  $self->{end} = $end-1;

  $self->{score} = $score;
  $self->{strand} = $strand;
  $self->{phase} = $phase;

  # Loading attributes
  my @attributes_tab = split(";",$attributes);
  foreach my $attr (@attributes_tab) {
    my ($k,$v);
    if($self->{format} =~ /gff3/i || $self->{format} =~ /gff$/i) {
      ($k,$v) = $attr =~ /(\S+)=(.*)/;
    } elsif ($self->{format} =~ /gtf/i){
      ($k,$v) = $attr =~ /(\S+)\s+"(.*)"/;
    }else{
	croak "Missing format argument (gff3,gtf) in CracTools::GFF::Annotation constructor";
    }
    if(defined $k && defined $v) {
      if($k eq "Parent") {
        my @parents = split(',',$v);
        $v = \@parents;
      }
      $self->{attributes}{$k} = $v;
    } #else {
    #  carp("Error parsing attribute $attr");
    #}
  }

}


sub chr {
  my $self = shift;
  my $chr = shift;
  if(defined $chr) {
    $self->{chr} = $chr;
  }
  return $self->{chr};
}


sub source {
  my $self = shift;
  my $source = shift;
  if(defined $source) {
    $self->{source} = $source;
  }
  return $self->{source};
}


sub feature {
  my $self = shift;
  my $feature = shift;
  if(defined $feature) {
    $self->{feature} = $feature;
  }
  return $self->{feature};
}


sub start {
  my $self = shift;
  my $start = shift;
  if(defined $start) {
    $self->{start} = $start;
  }
  return $self->{start};
}


sub end {
  my $self = shift;
  my $end = shift;
  if(defined $end) {
    $self->{end} = $end;
  }
  return $self->{end};
}


sub score {
  my $self = shift;
  my $score = shift;
  if(defined $score) {
    $self->{score} = $score;
  }
  return $self->{score};
}


sub strand {
  my $self = shift;
  my $strand = shift;
  if(defined $strand) {
    $self->{strand} = $strand;
  }
  return $self->{strand};
}


sub gffStrand {
  my $self = shift;
  return CracTools::Utils::convertStrand($self->{strand});
}


sub phase {
  my $self = shift;
  my $phase = shift;
  if(defined $phase) {
    $self->{phase} = $phase;
  }
  return $self->{phase};
}


sub parents {
  my $self = shift;
  if(defined $self->attribute('Parent')) {
    return $self->attribute('Parent');
  } else {
    return [];
  }
}


sub attribute {
  my $self = shift;
  my $key = shift;
  my $value = shift;
  if(defined $value) {
    return $self->{attributes}{$key} = $value;
  } elsif(defined $key) {
    return $self->{attributes}{$key};
  } else {
    return undef;
    #croak ("Missing attribute key to retreive attribute value");
  }
}

1;

__END__