File Coverage

blib/lib/Bio/Lite.pm

Criterion	Covered	Total	%
statement	78	94	82.9
branch	18	38	47.3
condition	5	12	41.6
subroutine	13	15	86.6
pod	7	7	100.0
total	121	166	72.8

package Bio::Lite;
{
  $Bio::Lite::DIST = 'Bio-Lite';
}
# ABSTRACT: Lightweight and fast module with a simplified API to ease scripting in bioinformatics
$Bio::Lite::VERSION = '0.003';
use Carp;

use Exporter;
our @ISA = qw(Exporter);
our @EXPORT = qw(seqFileIterator reverseComplemente convertStrand pairedEndSeqFileIterator gffFileIterator getReadingFileHandle getWritingFileHandle);
our @EXPORT_OK = qw();


sub reverseComplemente {
  my $seq = reverse shift;
  $seq =~ tr/ACGTacgt/TGCAtgca/;
  return $seq;
}


my %conversion_hash = ( '+' => 1, '-' => '-1', '1' => '+', '-1' => '-');
sub convertStrand($) {
  my $strand = shift;
  return $conversion_hash{$strand};
}


sub seqFileIterator {
  my ($file,$format) = @_;

  croak "Missing file in argument of seqFileIterator" if !defined $file;

  # Get file handle for $file
  my $fh = getReadingFileHandle($file);

  # Automatic file extension detection
  if(!defined $format) {
    if($file =~ /\.(fasta|fa)(\.|$)/) {
      $format = 'fasta';
    } elsif($file =~ /\.(fastq|fq)(\.|$)/) {
      $format = 'fastq';
    } else {
      croak "Undefined file extension";
    }
  } else {
    $format = lc $format;
  }

  # FASTA ITERATOR
  if ($format eq 'fasta') {
    # Read prev line for FASTA because we dont know the number
    # of line used for the sequence
    my $prev_line = <$fh>;
    chomp $prev_line;
    return sub {
      my ($name,$seq,$qual);
      if(defined $prev_line) {
        ($name) = $prev_line =~ />(.*)$/;
        $prev_line = <$fh>;
        # Until we find a new sequence identifier ">", we
        # concatenate the lines corresponding to the sequence
        while(defined $prev_line && $prev_line !~ /^>/) {
          chomp $prev_line;
          $seq .= $prev_line;
          $prev_line = <$fh>;
        }
        return {name => $name, seq => $seq, qual => $qual};
      } else {
        return undef;
      }
    };
  # FASTQ ITERATOR
  } elsif ($format eq 'fastq') {
    return sub {
      my ($name,$seq,$qual);
      ($name) = <$fh> =~ /@(.*)$/;
      if(defined $name) {
        $seq = <$fh>;
        chomp $seq;
        <$fh>; # skip second seq name (useless line)
        $qual = <$fh>;
        chomp $qual;
        return {name => $name, seq => $seq, qual => $qual};
      } else {
        return undef;
      }
    };
  } else {
    croak "Undefined file format";
  }
}


sub pairedEndSeqFileIterator {
  my($file1,$file2,$format) = @_;

  my $it1 = seqFileIterator($file1,$format);
  my $it2 = seqFileIterator($file2,$format);

  return sub {
    my $entry1 = $it1->();
    my $entry2 = $it2->();
    if(defined $entry1 && defined $entry2) {
      return { read1 => $entry1, read2 => $entry2 };
    } else {
      return undef;
    }
  };
}


sub gffFileIterator {
  my ($file,$format) = @_;

  croak "Missing arguments in gffFileIterator" if !defined $file || !defined $format;

  $format = lc $format;
  my $fh = getReadingFileHandle($file);

  my $attribute_split;

  if ($format eq 'gff3') {
    $attribute_split = sub {my $attr = shift; return $attr =~ /(\S+)=(.*)/;};
  } elsif ($format eq 'gtf' || $format eq 'gff2') {
    $attribute_split = sub {my $attr = shift; return $attr  =~ /(\S+)\s+"(.*)"/;};
  } else {
    croak "Undefined gff format";
  }

  my $line = <$fh>;
  while($line =~ /^#/) {
    $line = <$fh>;
  }

  return sub {
    if (defined $line) {
      my($chr,$source,$feature,$start,$end,$score,$strand,$frame,$attributes) = split("\t",$line);
      my @attributes_tab = split(";",$attributes);
      my %attributes_hash;
      foreach my $attr (@attributes_tab) {
        my ($k,$v) = $attribute_split->($attr);
        $attributes_hash{$k} = $v;
      }
      $line = <$fh>; # Get next line
      return { chr        => $chr,
        source     => $source,
        feature    => $feature,
        start      => $start,
        end        => $end,
        score      => $score,
        strand     => $strand,
        frame      => $frame,
        attributes => \%attributes_hash,
      };
    } else {
      return undef;
    }
  };
}


sub getReadingFileHandle {
  my $file = shift;
  my $fh;
  if($file =~ /\.gz$/) {
    open($fh,"gunzip -c $file |") or die ("Cannot open $file");
  } else {
    open($fh,"< $file") or die ("Cannot open $file");
  }
  return $fh;
}


sub getWritingFileHandle {
  my $file = shift;
  my $fh;
  if($file =~ /\.gz$/) {
    open($fh,"| gzip > $file") or die ("Cannot open $file");
  } else {
    open($fh,"> $file") or die ("Cannot open $file");
  }
  return $fh;
}


1;

__END__