Bio::DB::TFBS namespace has been moved to its own distribution named after itself

[bioperl-live.git] / Bio / Align / Utilities.pm

package Bio::Align::Utilities;
use strict;
use warnings;
use Carp;
use Bio::Root::Version;

use Exporter 'import';
our @EXPORT_OK = qw(
  aa_to_dna_aln
  bootstrap_replicates
  cat
  bootstrap_replicates_codons
  dna_to_aa_aln
  most_common_sequences
);
our %EXPORT_TAGS = (all => \@EXPORT_OK);

#
# BioPerl module for Bio::Align::Utilities
#
# Please direct questions and support issues to <bioperl-l@bioperl.org>
#
# Cared for by Jason Stajich <jason-at-bioperl.org> and Brian Osborne
#
# Copyright Jason Stajich
#
# You may distribute this module under the same terms as perl itself

# POD documentation - main docs before the code

=head1 NAME

Bio::Align::Utilities - A collection of utilities regarding converting
and manipulating alignment objects

=head1 SYNOPSIS

  use Bio::Align::Utilities qw(:all);

  # Even if the protein alignments are local make sure the start/end
  # stored in the LocatableSeq objects are to the full length protein.
  # The coding sequence that is passed in should still be the full 
  # length CDS as the nt alignment will be generated.
  # %dnaseqs is a hash of CDS sequences (spliced)
  my $dna_aln = aa_to_dna_aln($aa_aln,\%dnaseqs);

  # The reverse, which is simpler. The input alignment has to be
  # translate-able, with gap lengths and an overall length divisible by 3
  my $aa_aln = dna_to_aa_aln($dna_al);

  # Generate bootstraps
  my $replicates = bootstrap_replicates($aln,$count);

=head1 DESCRIPTION

This module contains utility methods for manipulating sequence
alignments (L<Bio::Align::AlignI>) objects.

The B<aa_to_dna_aln> utility is essentially the same as the B<mrtrans>
program by Bill Pearson available at
ftp://ftp.virginia.edu/pub/fasta/other/mrtrans.shar.  Of course this
is a pure-Perl implementation, but just to mention that if anything
seems odd you can check the alignments generated against Bill's
program.

=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 Support 

Please direct usage questions or support issues to the mailing list:

I<bioperl-l@bioperl.org>

rather than to the module maintainer directly. Many experienced and 
reponsive experts will be able look at the problem and quickly 
address it. Please include a thorough description of the problem 
with code and data examples if at all possible.

=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:

  https://github.com/bioperl/bioperl-live/issues

=head1 AUTHOR - Jason Stajich

Email jason-at-bioperl-dot-org

=head1 APPENDIX

The rest of the documentation details each of the object methods.
Internal methods are usually preceded with a _

=cut

use constant CODONSIZE => 3;
our $GAP       = '-';
our $CODONGAP  = $GAP x CODONSIZE;

=head2 aa_to_dna_aln

 Title   : aa_to_dna_aln
 Usage   : my $dnaaln = aa_to_dna_aln($aa_aln, \%seqs);
 Function: Will convert an AA alignment to DNA space given the 
           corresponding DNA sequences.  Note that this method expects 
           the DNA sequences to be in frame +1 (GFF frame 0) as it will
           start to project into coordinates starting at the first base of 
           the DNA sequence, if this alignment represents a different 
           frame for the cDNA you will need to edit the DNA sequences
           to remove the 1st or 2nd bases (and revcom if things should be).
 Returns : Bio::Align::AlignI object 
 Args    : 2 arguments, the alignment and a hashref.
           Alignment is a Bio::Align::AlignI of amino acid sequences. 
           The hash reference should have keys which are 
           the display_ids for the aa 
           sequences in the alignment and the values are a 
           Bio::PrimarySeqI object for the corresponding 
           spliced cDNA sequence. 

See also: L<Bio::Align::AlignI>, L<Bio::SimpleAlign>, L<Bio::PrimarySeq>

=cut

sub aa_to_dna_aln {
    my ( $aln, $dnaseqs ) = @_;
    unless ( defined $aln
        && ref($aln)
        && $aln->isa('Bio::Align::AlignI') )
    {
        croak(
'Must provide a valid Bio::Align::AlignI object as the first argument to aa_to_dna_aln, see the documentation for proper usage and the method signature'
        );
    }
    my $alnlen   = $aln->length;
    my $dnaalign = Bio::SimpleAlign->new();
    $aln->map_chars( '\.', $GAP );

    foreach my $seq ( $aln->each_seq ) {
        my $aa_seqstr = $seq->seq();
        my $pepid     = $seq->display_id;
        my $dnaseq    = $dnaseqs->{$pepid} || $aln->throw( "cannot find " . $seq->display_id );
        my $start_offset = ( $seq->start - 1 ) * CODONSIZE;
        $dnaseq = $dnaseq->seq();
        my $dnalen = $dnaseqs->{$pepid}->length;
        my $dnaid = $dnaseqs->{$pepid}->display_id || $pepid; # try to use DNAseq obj ID (issue #137)
        my $nt_seqstr;
        my $j = 0;
        for ( my $i = 0 ; $i < $alnlen ; $i++ ) {
            my $char = substr( $aa_seqstr, $i + $start_offset, 1 );
            if ( $char eq $GAP || $j >= $dnalen ) {
                $nt_seqstr .= $CODONGAP;
            }
            else {
                $nt_seqstr .= substr( $dnaseq, $j, CODONSIZE );
                $j += CODONSIZE;
            }
        }
        $nt_seqstr .= $GAP x ( ( $alnlen * 3 ) - length($nt_seqstr) );

        my $newdna = Bio::LocatableSeq->new(
            -display_id => $dnaid,
            -alphabet   => 'dna',
            -start      => $start_offset + 1,
            -end        => ( $seq->end * CODONSIZE ),
            -strand     => 1,
            -seq        => $nt_seqstr
        );
        $dnaalign->add_seq($newdna);
    }
    return $dnaalign;
}

=head2 dna_to_aa_aln

 Title   : dna_to_aa_aln
 Usage   : my $aa_aln = dna_to_aa_aln($dna_aln);
 Function: Convert a DNA alignment to an amino acid alignment where
           the length of all alignment strings and the lengths of any 
           gaps must be divisible by 3
 Returns : Bio::Align::AlignI object 
 Args    : the DNA alignment, a Bio::Align::AlignI of DNA sequences

See also: L<Bio::Align::AlignI>, L<Bio::SimpleAlign>, L<Bio::PrimarySeq>

=cut

sub dna_to_aa_aln {
    my $dna_aln = shift;
    unless ( defined $dna_aln
        && ref($dna_aln)
        && $dna_aln->isa('Bio::Align::AlignI') ) {
        croak(
'Must provide a valid Bio::Align::AlignI object as the argument to dna_to_aa_aln'
        );
    }
    my $codon_table = Bio::Tools::CodonTable->new;
    my $aa_aln      = Bio::SimpleAlign->new;

    for my $seq ( $dna_aln->each_seq ) {
        my ($aa_str, $aa_len);
        my @aln_str = split '', $seq->seq;
        croak("All lines in the alignment must have lengths divisible by 3")
          if ( scalar(@aln_str) % CODONSIZE );

        while ( @aln_str ) {
            my $triplet = join '', (splice( @aln_str, 0, CODONSIZE ));

            if ( $triplet =~ /^[GATC]+$/i ) {
                $aa_str .= $codon_table->translate($triplet);
                $aa_len++;
            }
            elsif ( $triplet =~ /^[$Bio::LocatableSeq::GAP_SYMBOLS]+$/ ) {
                $aa_str .= $GAP;
            }
            else {
                croak("The triplet '$triplet' is neither a valid codon nor all gaps");
            }
        }
        my $new_aa = Bio::LocatableSeq->new(
            -display_id => $seq->display_id,
            -alphabet   => 'protein',
            -start      => 1,
            -end        => $aa_len,
            -strand     => 1,
            -seq        => $aa_str
        );

        $aa_aln->add_seq($new_aa);
    }

    $aa_aln;
}

=head2 bootstrap_replicates

 Title   : bootstrap_replicates
 Usage   : my $alns = &bootstrap_replicates($aln,100);
 Function: Generate a pseudo-replicate of the data by randomly
           sampling, with replacement, the columns from an alignment for
           the non-parametric bootstrap.
 Returns : Arrayref of L<Bio::SimpleAlign> objects
 Args    : L<Bio::SimpleAlign> object
           Number of replicates to generate

=cut

sub bootstrap_replicates {
    my ( $aln, $count ) = @_;
    $count ||= 1;
    my $alen = $aln->length;
    my ( @seqs, @nm );
    $aln->set_displayname_flat(1);
    for my $s ( $aln->each_seq ) {
        push @seqs, $s->seq();
        push @nm,   $s->id;
    }
    my ( @alns, $i );
    while ( $count-- > 0 ) {
        my @newseqs;
        for ( $i = 0 ; $i < $alen ; $i++ ) {
            my $index = int( rand($alen) );
            my $c     = 0;
            for (@seqs) {
                $newseqs[ $c++ ] .= substr( $_, $index, 1 );
            }
        }
        my $newaln = Bio::SimpleAlign->new();
        my $i      = 0;
        for my $s (@newseqs) {
            ( my $tmp = $s ) =~ s/[$Bio::LocatableSeq::GAP_SYMBOLS]+//g;
            $newaln->add_seq(
                Bio::LocatableSeq->new(
                    -start      => 1,
                    -end        => length($tmp),
                    -display_id => $nm[ $i++ ],
                    -seq        => $s
                )
            );
        }
        push @alns, $newaln;
    }
    return \@alns;
}

=head2 bootstrap_replicates_codons

 Title   : bootstrap_replicates_codons
 Usage   : my $alns = &bootstrap_replicates_codons($aln,100);
 Function: Generate a pseudo-replicate of the data by randomly
           sampling, with replacement, the columns from a codon alignment for
           the non-parametric bootstrap. The alignment is assumed to start on
           the first position of a codon.
 Returns : Arrayref of L<Bio::SimpleAlign> objects
 Args    : L<Bio::SimpleAlign> object
           Number of replicates to generate

=cut

sub bootstrap_replicates_codons {
    my ( $aln, $count ) = @_;
    $count ||= 1;
    my $alen   = $aln->length;
    my $ncodon = int( $alen / 3 );
    my ( @seqs, @nm );
    $aln->set_displayname_flat(1);
    for my $s ( $aln->each_seq ) {
        push @seqs, $s->seq();
        push @nm,   $s->id;
    }
    my ( @alns, $i );
    while ( $count-- > 0 ) {
        my @newseqs;
        for ( $i = 0 ; $i < $ncodon ; $i++ ) {
            my $index  = int( rand($ncodon) );
            my $seqpos = $index * 3;
            my $c      = 0;
            for (@seqs) {
                $newseqs[ $c++ ] .= substr( $_, $seqpos, 3 );
            }
        }
        my $newaln = Bio::SimpleAlign->new();
        my $i      = 0;
        for my $s (@newseqs) {
            ( my $tmp = $s ) =~ s{[$Bio::LocatableSeq::GAP_SYMBOLS]+}{}g;
            $newaln->add_seq(
                Bio::LocatableSeq->new(
                    -start      => 1,
                    -end        => length($tmp),
                    -display_id => $nm[ $i++ ],
                    -seq        => $s
                )
            );
        }
        push @alns, $newaln;
    }
    return \@alns;
}

=head2 cat

 Title     : cat
 Usage     : $aln123 = cat($aln1, $aln2, $aln3)
 Function  : Concatenates alignment objects. Sequences are identified by id.
             An error will be thrown if the sequence ids are not unique in the
             first alignment. If any ids are not present or not unique in any
             of the additional alignments then those sequences are omitted from
             the concatenated alignment, and a warning is issued. An error will
             be thrown if any of the alignments are not flush, since
             concatenating such alignments is unlikely to make biological
             sense.
 Returns   : A new Bio::SimpleAlign object
 Args      : A list of Bio::SimpleAlign objects

=cut

sub cat {
    my ( $self, @aln ) = @_;
    $self->throw("cat method called with no arguments") unless $self;
    for ( $self, @aln ) {
        $self->throw( $_->id . " is not a Bio::Align::AlignI object" )
          unless $_->isa('Bio::Align::AlignI');
        $self->throw( $_->id . " is not flush" ) unless $_->is_flush;
    }
    my $aln = $self->new;
    $aln->id( $self->id );
    $aln->annotation( $self->annotation );
    my %unique;
  SEQ: foreach my $seq ( $self->each_seq() ) {
        throw( "ID: ", $seq->id, " is not unique in initial alignment." )
          if exists $unique{ $seq->id };
        $unique{ $seq->id } = 1;

        # Can be Bio::LocatableSeq, Bio::Seq::Meta or Bio::Seq::Meta::Array
        my $new_seq = $seq->new(
            -id      => $seq->id,
            -strand  => $seq->strand,
            -verbose => $self->verbose
        );
        $new_seq->seq( $seq->seq );
        $new_seq->start( $seq->start );
        $new_seq->end( $seq->end );
        if ( $new_seq->isa('Bio::Seq::MetaI') ) {
            for my $meta_name ( $seq->meta_names ) {
                $new_seq->named_submeta( $meta_name, $new_seq->start,
                    $new_seq->end, $seq->named_meta($meta_name) );
            }
        }
        for my $cat_aln (@aln) {
            my @cat_seq = $cat_aln->each_seq_with_id( $seq->id );
            if ( @cat_seq == 0 ) {
                $self->warn( $seq->id
                      . " not found in alignment "
                      . $cat_aln->id
                      . ", skipping this sequence." );
                next SEQ;
            }
            if ( @cat_seq > 1 ) {
                $self->warn( $seq->id
                      . " found multiple times in alignment "
                      . $cat_aln->id
                      . ", skipping this sequence." );
                next SEQ;
            }
            my $cat_seq = $cat_seq[0];
            my $old_end = $new_seq->end;
            $new_seq->seq( $new_seq->seq . $cat_seq->seq );

            # Not sure if this is a sensible way to deal with end coordinates
            $new_seq->end(
                $new_seq->end + $cat_seq->end + 1 - $cat_seq->start );

            if ( $cat_seq->isa('Bio::Seq::Meta::Array') ) {
                unless ( $new_seq->isa('Bio::Seq::Meta::Array') ) {
                    my $meta_seq = Bio::Seq::Meta::Array->new;
                    $meta_seq->seq( $new_seq->seq );
                    $meta_seq->start( $new_seq->start );
                    $meta_seq->end( $new_seq->end );
                    if ( $new_seq->isa('Bio::Seq::Meta') ) {
                        for my $meta_name ( $new_seq->meta_names ) {
                            $meta_seq->named_submeta(
                                $meta_name,
                                $new_seq->start,
                                $old_end,
                                [
                                    split(
                                        //, $new_seq->named_meta($meta_name)
                                    )
                                ]
                            );
                        }
                    }
                    $new_seq = $meta_seq;
                }
                for my $meta_name ( $cat_seq->meta_names ) {
                    $new_seq->named_submeta( $meta_name, $old_end + 1,
                        $new_seq->end, $cat_seq->named_meta($meta_name) );
                }
            }
            elsif ( $cat_seq->isa('Bio::Seq::Meta') ) {
                if ( $new_seq->isa('Bio::Seq::Meta::Array') ) {
                    for my $meta_name ( $cat_seq->meta_names ) {
                        $new_seq->named_submeta( $meta_name, $old_end + 1,
                            $new_seq->end,
                            [ split( //, $cat_seq->named_meta($meta_name) ) ] );
                    }
                }
                else {
                    unless ( $new_seq->isa('Bio::Seq::Meta') ) {
                        my $meta_seq = Bio::Seq::Meta::Array->new;
                        $meta_seq->seq( $new_seq->seq );
                        $meta_seq->start( $new_seq->start );
                        $meta_seq->end( $new_seq->end );
                        $new_seq = $meta_seq;
                    }
                    for my $meta_name ( $cat_seq->meta_names ) {
                        $new_seq->named_submeta( $meta_name, $old_end + 1,
                            $new_seq->end, $cat_seq->named_meta($meta_name) );
                    }
                }
            }
        }
        $aln->add_seq($new_seq);
    }
    my $cons_meta = $self->consensus_meta;
    my $new_cons_meta;
    if ($cons_meta) {
        $new_cons_meta = Bio::Seq::Meta->new();
        for my $meta_name ( $cons_meta->meta_names ) {
            $new_cons_meta->named_submeta( $meta_name, 1, $self->length,
                $cons_meta->$meta_name );
        }
    }
    my $end = $self->length;
    for my $cat_aln (@aln) {
        my $cat_cons_meta = $cat_aln->consensus_meta;
        if ($cat_cons_meta) {
            $new_cons_meta = Bio::Seq::Meta->new() if !$new_cons_meta;
            for my $meta_name ( $cat_cons_meta->meta_names ) {
                $new_cons_meta->named_submeta(
                    $meta_name, $end + 1,
                    $end + $cat_aln->length,
                    $cat_cons_meta->$meta_name
                );
            }
        }
        $end += $cat_aln->length;
    }
    $aln->consensus_meta($new_cons_meta) if $new_cons_meta;
    return $aln;
}


=head2 most_common_sequences

 Title     : most_common_sequences
 Usage     : @common = most_common_sequences ($align, $case_sensitivity)
 Function  : Returns an array of the sequences that appear most often in the
             alignment (although this probably makes more sense when there is
             only a single most common sequence).  Sequences are compared after
             removing any "-" (gap characters), and ambiguous units (e.g., R
             for purines) are only compared to themselves.  The returned
             sequence is also missing the "-" since they don't actually make
             part of the sequence.
 Returns   : Array of text strings.
 Arguments : Optional argument defining whether the comparison between sequences
             to find the most common should be case sensitive. Defaults to
             false, i.e, not case sensitive.

=cut

sub most_common_sequences {
    my $align = shift
      or croak ("Must provide Bio::AlignI object to Bio::Align::Utilities::most_common_sequences");
    my $case_sensitive = shift; # defaults to false (we get undef if nothing)

    ## We keep track of the max on this loop. Saves us having to
    ## transverse the hash table later to find the maximum value.
    my $max = 0;
    my %counts;
    foreach ($align->each_seq) {
        (my $seq = $_->seq) =~ tr/-//d;
        $seq = uc ($seq) unless $case_sensitive;
        $max++ if (++$counts{$seq} > $max);
    }
    my @common = grep ($counts{$_} == $max, keys %counts);
    return @common;
}

1;