—
—
—
—
—
—
—

              
# $Id: phase.pm,v 1.7.4.1 2006/10/02 23:10:23 sendu Exp $
#
# BioPerl module for Bio::PopGen::IO::phase
#
# Cared for by Rich Dobson <r.j.dobson-at-qmul.ac.uk>
#
# Copyright Rich Dobson
#
# You may distribute this module under the same terms as perl itself
# POD documentation - main docs before the code
HideShow 60 lines of Pod
=head1 NAME
Bio::PopGen::IO::phase - A parser for Phase format data
=head1 SYNOPSIS
# Do not use directly, use through the Bio::PopGen::IO driver
  use Bio::PopGen::IO;
  my $io = new Bio::PopGen::IO(-format => 'phase',
                               -file   => 'data.phase');
  # Some IO might support reading in a population at a time
  my @population;
  while( my $ind = $io->next_individual ) {
      push @population, $ind;
  }
=head1 DESCRIPTION
A driver module for Bio::PopGen::IO for parsing phase data.
PHASE is defined in http://www.stat.washington.edu/stephens/instruct2.1.pdf
=head1 FEEDBACK
=head2 Mailing Lists
User feedback is an integral part of the evolution of this and other
Bioperl modules. Send your comments and suggestions preferably to
the Bioperl mailing list.  Your participation is much appreciated.
  bioperl-l@bioperl.org                  - General discussion
  http://bioperl.org/wiki/Mailing_lists  - About the mailing lists
=head2 Reporting Bugs
Report bugs to the Bioperl bug tracking system to help us keep track
of the bugs and their resolution. Bug reports can be submitted via
the web:
  http://bugzilla.open-bio.org/
=head1 AUTHOR - Rich Dobson
Email r.j.dobson-at-qmul.ac.uk
=head1 CONTRIBUTORS
Jason Stajich, jason-at-bioperl.org
=head1 APPENDIX
The rest of the documentation details each of the object methods.
Internal methods are usually preceded with a _
=cut
# Let the code begin...
package Bio::PopGen::IO::phase;
use vars qw($FieldDelim $AlleleDelim $NoHeader);
use strict;
($FieldDelim,$AlleleDelim,$NoHeader) =( ',', '\s+',0);
use Bio::PopGen::Individual;
use Bio::PopGen::Population;
use Bio::PopGen::Genotype;
use base qw(Bio::PopGen::IO);
HideShow 14 lines of Pod
=head2 new
 Title   : new
 Usage   : my $obj = new Bio::PopGen::IO::hapmap();
 Function: Builds a new Bio::PopGen::IO::hapmap object 
 Returns : an instance of Bio::PopGen::IO::hapmap
 Args    : [optional, these are the current defaults] 
           -field_delimiter => ','
           -allele_delimiter=> '\s+'
           -no_header       => 0,
=cut
sub _initialize  {
    my($self, @args) = @_;
    $Bio::PopGen::Genotype::BlankAlleles='';
    my ($fieldsep,$all_sep, 
        $noheader) = $self->_rearrange([qw(FIELD_DELIMITER
                                           ALLELE_DELIMITER
                                           NO_HEADER)],@args);
    $self->flag('no_header', defined $noheader ? $noheader : $NoHeader);
    $self->flag('field_delimiter',defined $fieldsep ? $fieldsep : $FieldDelim);
    $self->flag('allele_delimiter',defined $all_sep ? $all_sep : $AlleleDelim);
    $self->{'_header'} = undef;
    return 1;
}
HideShow 13 lines of Pod
=head2 flag
 Title   : flag
 Usage   : $obj->flag($flagname,$newval)
 Function: Get/Set the flag value
 Returns : value of a flag (a boolean)
 Args    : A flag name, currently we expect 
           'no_header', 'field_delimiter', or 'allele_delimiter' 
           on set, new value (a boolean or undef, optional)
=cut
sub flag  {
    my $self = shift;
    my $fieldname = shift;
    return unless defined $fieldname;
    return $self->{'_flag'}->{$fieldname} = shift if @_;
    return $self->{'_flag'}->{$fieldname};
}
HideShow 11 lines of Pod
=head2 next_individual
 Title   : next_individual
 Usage   : my $ind = $popgenio->next_individual;
 Function: Retrieve the next individual from a dataset
 Returns : L<Bio::PopGen::IndividualI> object
 Args    : none
=cut
sub next_individual  {
    my ($self) = @_;
    my ($sam,@marker_results,$number_of_ids,$number_of_markers,
        $marker_positions,$micro_snp);
    while( defined( $_ = $self->_readline) ) {
        next if( /^\s+$/ || ! length($_) );
        last;
    }
     
    return unless defined $_; 
    if( $self->flag('no_header') || defined $self->{'_header'} ) {
        ####### sometimes there is some marker info at the start of a phase input file
        ####### we collect it in the next few lines if there is. Should this info be held in a marker object?
        if(!$self->{'_count'} && /^\s*\d+$/){
            $self->flag('number_of_ids',$_);
            #print "number_of_ids : $number_of_ids\n";
            $self->{'_count'}++;
            return $self->next_individual;
        } elsif($self->{'_count'} == 1 && /^\s*\d+$/){
            $self->flag('number_of_markers',$_);
            #print "number_of_markers : $number_of_markers\n";
            $self->{'_count'}++;
            return $self->next_individual;
        } elsif($self->{'_count'} == 2 && /^\s*P\s\d/){
            $self->flag('marker_positions',$_);
            #print "marker_position : $marker_positions\n";
            $self->{'_count'}++;
            return $self->next_individual;
        } elsif($self->{'_count'} == 3 && /^\s*(M|S)+\s*$/i){
            $self->flag('micro_snp',$_);
            #print "microsat or snp : $micro_snp\n";
            $self->{'_count'}++;
            return $self->next_individual;
        } elsif(/^\s*\#/){
            ($self->{'_sam'}) = /^\s*\#(.+)/;
            #print "sample : $self->{'_sam'}\n";
            $self->{'_count'}++;
            return $self->next_individual;
        } else {
            chomp $_;
            if( $self->{'_row1'} ) {
                # if we are looking at the 2nd row of alleles for this id
                @{$self->{'_second_row'}} = 
                    split($self->flag('field_delimiter'),$_);
                for my $i(0 .. $#{$self->{'_first_row'}}){
                    push(@{$self->{'_marker_results'}},
                         $self->{'_first_row'}->[$i].
                         $self->flag('field_delimiter').
                         $self->{'_second_row'}->[$i]);
                }
                $self->{'_row1'} = 0;
            } else {
                # if we are looking at the first row of alleles for this id
                @{$self->{'_marker_results'}} = ();
                @{$self->{'_first_row'}} = split($self->flag('field_delimiter'),$_);
                $self->{'_row1'} = 1;
                return $self->next_individual;
            }
        }
        my $i = 1;
        foreach my $m ( @{$self->{'_marker_results'}} ) {
            $m =~ s/^\s+//;
            $m =~ s/\s+$//;
            my $markername;
            if( defined $self->{'_header'} ) {
                $markername = $self->{'_header'}->[$i] || "Marker$i";
            } else { 
                $markername = "Marker$i";
            }
            $self->debug( "markername is $markername alleles are $m\n");
            my @alleles = split($self->flag('allele_delimiter'), $m);        
            $m = new Bio::PopGen::Genotype(-alleles      =>\@alleles,
                                           -marker_name  => $markername,
                                           -individual_id=> $self->{'_sam'}); 
            $i++; 
        }
        return new Bio::PopGen::Individual(-unique_id => $self->{'_sam'},
                                           -genotypes =>\@{$self->{'_marker_results'}},
                                           );
    } else {
        chomp;
        $self->{'_header'} = [split($self->flag('field_delimiter'),$_)];
        return $self->next_individual; # rerun loop again
    }
    return;
}
HideShow 11 lines of Pod
=head2 next_population
 Title   : next_population
 Usage   : my $ind = $popgenio->next_population;
 Function: Retrieve the next population from a dataset
 Returns : L<Bio::PopGen::PopulationI> object
 Args    : none
 Note    : Many implementation will not implement this
=cut
sub next_population{
    my ($self) = @_;
    my @inds;
    while( my $ind = $self->next_individual ) {
        push @inds, $ind;
    }
    Bio::PopGen::Population->new(-individuals => \@inds);
}
HideShow 10 lines of Pod
=head2 write_individual
 Title   : write_individual
 Usage   : $popgenio->write_individual($ind);
 Function: Write an individual out in the file format
 Returns : none
 Args    : L<Bio::PopGen::PopulationI> object(s)
=cut
sub write_individual {
    my ($self,@inds) = @_;
    my $fielddelim  = $self->flag('field_delimiter');
    my $alleledelim = $self->flag('allele_delimiter');
    foreach my $ind ( @inds ) {
        if (! ref($ind) || ! $ind->isa('Bio::PopGen::IndividualI') ) {
            $self->warn("Cannot write an object that is not a Bio::PopGen::IndividualI object ($ind)");
            next;
        }
        # we'll go ahead and sort these until
        # we have a better way to insure a consistent order
        my @marker_names = sort $ind->get_marker_names;
        if( ! $self->flag('no_header') && 
            ! $self->flag('header_written') ) {
            $self->_print(join($fielddelim, ('SAM', @marker_names)), "\n");
            $self->flag('header_written',1);
        }
        my(@row1,@row2);
        for (@marker_names){
            my $geno = $ind->get_Genotypes(-marker => $_);
            my @alleles = $geno->get_Alleles();
            push(@row1,$alleles[0]);
            push(@row2,$alleles[1]);
        }
        $self->_print("#",$ind->unique_id,"\n",
                      join($fielddelim,@row1),"\n",
                      join($fielddelim,@row2),"\n");
    }
}
HideShow 11 lines of Pod
=head2 write_population
 Title   : write_population
 Usage   : $popgenio->write_population($pop);
 Function: Write a population out in the file format
 Returns : none
 Args    : L<Bio::PopGen::PopulationI> object(s)
 Note    : Many implementation will not implement this
=cut
sub write_population {
    my ($self,@pops) = @_;
    my $fielddelim  = $self->flag('field_delimiter');
    my $alleledelim = $self->flag('allele_delimiter');
    foreach my $pop ( @pops ) {
        if (! ref($pop) || ! $pop->isa('Bio::PopGen::PopulationI') ) {
            $self->warn("Cannot write an object that is not a Bio::PopGen::PopulationI object");
            next;
        }
        # we'll go ahead and sort these until
        # we have a better way to insure a consistent order
        my @marker_names = sort $pop->get_marker_names;
        if( ! $self->flag('no_header') && 
            ! $self->flag('header_written') ) {
            $self->_print( join($fielddelim, ('SAM', @marker_names)), 
                           "\n");
            $self->flag('header_written',1);
        }
        foreach my $ind ( $pop->get_Individuals ) {
            my(@row1,@row2);
            for (@marker_names){
                my $geno = $ind->get_Genotypes(-marker => $_);
                my @alleles = $geno->get_Alleles();
                push (@row1,$alleles[0]);
                push (@row2,$alleles[1]);
            }
            $self->_print("#",$ind->unique_id,"\n",
                          join($fielddelim,@row1),"\n",
                          join($fielddelim,@row2),"\n");
        } 
    }
}
1;