# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>> DESCRIPTION <<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
-# yak yak yak
+# Routines to create a pair-wise alignment.
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
use warnings;
use strict;
use Data::Dumper;
-use Storable qw( dclone );
use Maasha::Calc;
use Maasha::Seq;
use vars qw ( @ISA @EXPORT );
+@ISA = qw( Exporter );
+
use constant {
- Q_BEG => 0,
- Q_END => 1,
- S_BEG => 2,
- S_END => 3,
- LEN => 4,
- SCORE => 5,
- HEAD => 0,
- SEQ => 1,
- ALPH_LEN => 4,
+ SCORE_FACTOR_LEN => 3,
+ SCORE_FACTOR_CORNER => 2,
+ SCORE_FACTOR_DIAG => 1,
+ SCORE_FACTOR_NARROW => 0.5,
+ VERBOSE => 0,
};
-@ISA = qw( Exporter );
-
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<
# Generates an alignment by chaining matches, which are subsequences
# shared between two sequences. The routine functions by considering
# only matches within a given search space. If no matches are given
- # these will be generated, if long matches are found these will be
- # included in the alignment, otherwise matches will be included depending
- # on a calculated score. New search spaces spanning the spaces between
- # matches and the search space boundaries will be cast and recursed into.
+ # these will be located and matches will be included depending on a
+ # calculated score. New search spaces spanning the spaces between
+ # matches and the search space boundaries will be cast and recursed
+ # into.
- my ( $q_seq, # sequence 1 ref
- $s_seq, # sequence 2 ref
+ # Usage: align_two_seq( { Q_SEQ => \$q_seq, S_SEQ => \$s_seq }, [] );
+
+ my ( $space, # search space
$matches, # list of matches
- $q_min, # q sequence start position
- $q_max, # q sequecne stop position
- $s_min, # s sequence start position
- $s_max, # s sequecne stop position
) = @_;
- # returns a chain of matches that can be chained into an alignment
+ # Returns a list.
- $matches ||= [];
- $q_min ||= 0;
- $s_min ||= 0;
- $q_max ||= length( ${ $q_seq } ) - 1;
- $s_max ||= length( ${ $s_seq } ) - 1;
+ my ( @chain, $best_match, $left_space, $right_space );
- my ( $wordsize, @chain, $match, $best_match, @long_matches );
+ new_space( $space );
- $matches = select_matches( $matches, $q_min, $q_max, $s_min, $s_max );
+ matches_select( $matches, $space );
if ( scalar @{ $matches } == 0 ) # no matches - find some!
{
- $wordsize = find_wordsize( $q_min, $q_max, $s_min, $s_max );
- $matches = find_matches( $q_seq, $s_seq, $wordsize, $q_min, $q_max, $s_min, $s_max );
+ $matches = matches_find( $space );
- while ( scalar @{ $matches } == 0 and $wordsize > 1 )
- {
- $wordsize--;
- $matches = find_matches( $q_seq, $s_seq, $wordsize, $q_min, $q_max, $s_min, $s_max );
- }
-
if ( scalar @{ $matches } > 0 ) {
- push @chain, align_two_seq( $q_seq, $s_seq, $matches, $q_min, $q_max, $s_min, $s_max );
+ push @chain, align_two_seq( $space, $matches );
}
}
else # matches are included according to score
{
- foreach $match ( @{ $matches } ) {
- $match->[ SCORE ] = score_match( $match, $q_min, $q_max, $s_min, $s_max );
- }
+ $matches = matches_filter( $space, $matches );
- @{ $matches } = grep { $_->[ SCORE ] > 0 } @{ $matches };
- @{ $matches } = sort { $b->[ SCORE ] <=> $a->[ SCORE ] } @{ $matches };
-
- $best_match = shift @{ $matches };
+ $best_match = shift @{ $matches };
if ( $best_match )
{
push @chain, $best_match;
- if ( $best_match->[ Q_BEG ] - $q_min >= 2 and $best_match->[ S_BEG ] - $s_min >= 2 ) {
- push @chain, align_two_seq( $q_seq, $s_seq, $matches, $q_min, $best_match->[ Q_BEG ] - 1, $s_min, $best_match->[ S_BEG ] - 1 ); # left search space
- }
+ $left_space = new_space_left( $best_match, $space );
+ $right_space = new_space_right( $best_match, $space );
- if ( $q_max - $best_match->[ Q_END ] >= 2 and $s_max - $best_match->[ S_END ] >= 2 ) {
- push @chain, align_two_seq( $q_seq, $s_seq, $matches, $best_match->[ Q_END ] + 1, $q_max, $best_match->[ S_END ] + 1, $s_max ); # right search space
- }
+ push @chain, align_two_seq( $left_space, $matches ) if defined $left_space;
+ push @chain, align_two_seq( $right_space, $matches ) if defined $right_space;
}
}
}
-sub select_matches
+sub new_space
+{
+ # Martin A. Hansen, August 2009.
+
+ # Define a new search space if not previously
+ # defined. This occurs if align_two_seq() is
+ # called initially with only the sequences.
+
+ my ( $space, # search space
+ ) = @_;
+
+ # Returns nothing.
+
+ if ( not defined $space->{ 'Q_MAX' } )
+ {
+ $space->{ 'Q_MIN' } = 0;
+ $space->{ 'S_MIN' } = 0;
+ $space->{ 'Q_MAX' } = length( ${ $space->{ 'Q_SEQ' } } ) - 1,
+ $space->{ 'S_MAX' } = length( ${ $space->{ 'S_SEQ' } } ) - 1,
+ }
+}
+
+
+sub new_space_left
+{
+ # Martin A. Hansen, August 2009.
+
+ # Create a new search space between the beginning of
+ # the current search space and the best match.
+
+ my ( $best_match, # best match
+ $space, # search space
+ ) = @_;
+
+ # Returns hashref.
+
+ my ( $new_space );
+
+ if ( $best_match->{ 'Q_BEG' } - $space->{ 'Q_MIN' } >= 2 and $best_match->{ 'S_BEG' } - $space->{ 'S_MIN' } >= 2 )
+ {
+ $new_space = {
+ Q_SEQ => $space->{ 'Q_SEQ' },
+ S_SEQ => $space->{ 'S_SEQ' },
+ Q_MIN => $space->{ 'Q_MIN' },
+ Q_MAX => $best_match->{ 'Q_BEG' } - 1,
+ S_MIN => $space->{ 'S_MIN' },
+ S_MAX => $best_match->{ 'S_BEG' } - 1,
+ };
+ }
+
+ return wantarray ? %{ $new_space } : $new_space;
+}
+
+
+sub new_space_right
+{
+ # Martin A. Hansen, August 2009.
+
+ # Create a new search space between the best match
+ # and the end of the current search space.
+
+ my ( $best_match, # best match
+ $space, # search space
+ ) = @_;
+
+ # Returns hashref.
+
+ my ( $new_space );
+
+ if ( $space->{ 'Q_MAX' } - $best_match->{ 'Q_END' } >= 2 and $space->{ 'S_MAX' } - $best_match->{ 'S_END' } >= 2 )
+ {
+ $new_space = {
+ Q_SEQ => $space->{ 'Q_SEQ' },
+ S_SEQ => $space->{ 'S_SEQ' },
+ Q_MIN => $best_match->{ 'Q_END' } + 1,
+ Q_MAX => $space->{ 'Q_MAX' },
+ S_MIN => $best_match->{ 'S_END' } + 1,
+ S_MAX => $space->{ 'S_MAX' },
+ };
+ }
+
+ return wantarray ? %{ $new_space } : $new_space;
+}
+
+
+sub matches_select
{
# Martin A. Hansen, August 2006.
# this search space.
my ( $matches, # list of matches
- $q_min, # q sequence start position
- $q_max, # q sequecne stop position
- $s_min, # s sequence start position
- $s_max, # s sequecne stop position
+ $space, # search space
) = @_;
- # returns list of matches
-
- my ( @matches );
-
- @matches = grep { $_->[ Q_BEG ] >= $q_min and
- $_->[ S_BEG ] >= $s_min and
- $_->[ Q_END ] <= $q_max and
- $_->[ S_END ] <= $s_max } @{ $matches };
+ # Returns nothing.
- return wantarray ? @matches : \@matches;
+ @{ $matches } = grep { $_->{ 'Q_BEG' } >= $space->{ 'Q_MIN' } and
+ $_->{ 'S_BEG' } >= $space->{ 'S_MIN' } and
+ $_->{ 'Q_END' } <= $space->{ 'Q_MAX' } and
+ $_->{ 'S_END' } <= $space->{ 'S_MAX' } } @{ $matches };
}
-sub find_wordsize
+sub word_size_calc
{
- # Martin A. Hansen, August 2006.
-
+ # Martin A. Hansen, August 2008.
+
# Given a search space calculates the wordsize for a word so a match
# occurs only a few times. More matches may be needed at low similarity in
# order to avoid starting with a wrong match.
- my ( $q_min, # q sequence start position
- $q_max, # q sequecne stop position
- $s_min, # s sequence start position
- $s_max, # s sequecne stop position
+ my ( $space, # search space
) = @_;
- # returns integer
+ # Returns integer.
- my ( $q_dim, $s_dim, $dim_min, $wordsize );
+ my ( $factor, $word_size );
- $q_dim = $q_max - $q_min + 1;
- $s_dim = $s_max - $s_min + 1;
+ $factor = 0.05;
- $dim_min = Maasha::Calc::min( $q_dim, $s_dim );
+ if ( $space->{ 'Q_MAX' } - $space->{ 'Q_MIN' } < $space->{ 'S_MAX' } - $space->{ 'S_MIN' } ) {
+ $word_size = int( $factor * ( $space->{ 'Q_MAX' } - $space->{ 'Q_MIN' } ) ) + 1;
+ } else {
+ $word_size = int( $factor * ( $space->{ 'S_MAX' } - $space->{ 'S_MIN' } ) ) + 1;
+ }
- $wordsize = 1;
+ return $word_size;
+}
- if ( $dim_min > 2000000 ) # optimized for DNA
- {
- $wordsize = 50;
- }
- elsif ( $dim_min > 100000 ) # optimized for DNA
- {
- $wordsize = 25;
- }
- elsif ( $q_dim > 100 or $s_dim > 100 ) # optimized for DNA
+
+sub seq_index
+{
+ # Martin A. Hansen, August 2009.
+
+ # Indexes a query sequence in a specified search space
+ # Overlapping words are saved as keys in an index hash,
+ # and a list of word postions are saved as value.
+
+ my ( $space, # search space
+ $word_size, # word size
+ ) = @_;
+
+ # Returns a hashref.
+
+ my ( $i, $word, %index );
+
+ for ( $i = $space->{ 'Q_MIN' }; $i <= $space->{ 'Q_MAX' } - $word_size + 1; $i++ )
{
- while ( ALPH_LEN ** $wordsize <= $q_dim * $s_dim and $wordsize < $dim_min ) {
- $wordsize++;
- }
+ $word = uc substr ${ $space->{ 'Q_SEQ' } }, $i, $word_size;
+
+ next if $word =~ tr/N//; # Skip sequences with Ns.
+
+ push @{ $index{ $word } }, $i;
}
- else
+
+ return wantarray ? %index : \%index;
+}
+
+
+sub seq_scan
+{
+ # Martin A. Hansen, August 2009.
+
+ my ( $index, # sequence index
+ $space, # search space
+ $word_size, # word size
+ ) = @_;
+
+ # Returns a list.
+
+ my ( $i, $word, $pos, $match, @matches, $redundant );
+
+ $redundant = {};
+
+ for ( $i = $space->{ 'S_MIN' }; $i <= $space->{ 'S_MAX' } - $word_size + 1; $i++ )
{
- while ( ALPH_LEN ** $wordsize <= $dim_min and $wordsize < $dim_min ) {
- $wordsize++;
+ $word = uc substr ${ $space->{ 'S_SEQ' } }, $i, $word_size;
+
+ if ( exists $index->{ $word } )
+ {
+ foreach $pos ( @{ $index->{ $word } } )
+ {
+ $match = {
+ 'Q_BEG' => $pos,
+ 'Q_END' => $pos + $word_size - 1,
+ 'S_BEG' => $i,
+ 'S_END' => $i + $word_size - 1,
+ 'LEN' => $word_size,
+ 'SCORE' => 0,
+ };
+
+ if ( not match_redundant( $match, $redundant ) )
+ {
+ match_expand( $match, $space );
+
+ push @matches, $match;
+
+ match_redundant_add( $match, $redundant );
+ }
+ }
}
}
- return $wordsize;
+ return wantarray ? @matches : \@matches;
}
-sub find_matches
+sub matches_find
{
# Martin A. Hansen, November 2006
- # given two sequences, find all maximum expanded matches between these
+ # Given two sequences, find all maximum expanded matches between these.
- my ( $q_seq, # sequence 1
- $s_seq, # sequence 2
- $wordsize, # word size
- $q_min, # q sequence start position
- $q_max, # q sequecne stop position
- $s_min, # s sequence start position
- $s_max, # s sequecne stop position
+ my ( $space, # search space
) = @_;
# returns list of matches
- my ( $q_beg, $q_word, %word_hash, $s_beg, $s_word, $match, @matches );
+ my ( $word_size, $index, $matches );
- if ( length ${ $s_seq } > length ${ $q_seq } )
+ $word_size = word_size_calc( $space );
+
+ $matches = [];
+
+ while ( scalar @{ $matches } == 0 and $word_size > 0 )
{
- for ( $q_beg = $q_min; $q_beg <= $q_max - $wordsize + 1; $q_beg++ )
- {
- $q_word = lc substr ${ $q_seq }, $q_beg, $wordsize;
+ $index = seq_index( $space, $word_size );
+ $matches = seq_scan( $index, $space, $word_size );
- next if $q_word =~ /n/i; # DNA/genome optimization
+ $word_size--;
+ }
- push @{ $word_hash{ $q_word } }, $q_beg;
- }
+ return wantarray ? @{ $matches } : $matches;
+}
- for ( $s_beg = $s_min; $s_beg <= $s_max - $wordsize + 1; $s_beg++ )
- {
- $s_word = lc substr ${ $s_seq }, $s_beg, $wordsize;
- if ( exists $word_hash{ $s_word } )
- {
- foreach $q_beg ( @{ $word_hash{ $s_word } } )
- {
- $match = [ $q_beg, $q_beg + $wordsize - 1, $s_beg, $s_beg + $wordsize - 1 ];
-
- if ( grep { $match->[ Q_BEG ] >= $_->[ Q_BEG ] and
- $match->[ Q_END ] <= $_->[ Q_END ] and
- $match->[ S_BEG ] >= $_->[ S_BEG ] and
- $match->[ S_END ] <= $_->[ S_END ] } @matches )
- {
- next; # match is redundant
- }
- else
- {
- $match = expand_match( $q_seq, $s_seq, $match, $q_max, $q_min, $s_max, $s_min );
- $match->[ LEN ] = $match->[ Q_END ] - $match->[ Q_BEG ] + 1;
-
- push @matches, $match;
- }
- }
- }
- }
- }
- else
- {
- for ( $s_beg = $s_min; $s_beg <= $s_max - $wordsize + 1; $s_beg++ )
- {
- $s_word = lc substr ${ $s_seq }, $s_beg, $wordsize;
+sub match_expand
+{
+ # Martin A. Hansen, August 2009.
- next if $s_word =~ /n/i; # DNA/genome optimization
+ # Given a match and a search space hash,
+ # expand the match maximally both forward and
+ # backward direction.
- push @{ $word_hash{ $s_word } }, $s_beg;
- }
+ my ( $match, # match to expand
+ $space, # search space
+ ) = @_;
- for ( $q_beg = $q_min; $q_beg <= $q_max - $wordsize + 1; $q_beg++ )
- {
- $q_word = lc substr ${ $q_seq }, $q_beg, $wordsize;
+ # Returns nothing.
- if ( exists $word_hash{ $q_word } )
- {
- foreach $s_beg ( @{ $word_hash{ $q_word } } )
- {
- $match = [ $q_beg, $q_beg + $wordsize - 1, $s_beg, $s_beg + $wordsize - 1 ];
-
- if ( grep { $match->[ Q_BEG ] >= $_->[ Q_BEG ] and
- $match->[ Q_END ] <= $_->[ Q_END ] and
- $match->[ S_BEG ] >= $_->[ S_BEG ] and
- $match->[ S_END ] <= $_->[ S_END ] } @matches )
- {
- next; # match is redundant
- }
- else
- {
- $match = expand_match( $q_seq, $s_seq, $match, $q_max, $q_min, $s_max, $s_min );
- $match->[ LEN ] = $match->[ Q_END ] - $match->[ Q_BEG ] + 1;
-
- push @matches, $match;
- }
- }
- }
- }
+ match_expand_forward( $match, $space );
+ match_expand_backward( $match, $space );
+}
+
+
+sub match_expand_forward
+{
+ # Martin A. Hansen, August 2009.
+
+ # Given a match and a search space hash,
+ # expand the match maximally in the forward
+ # direction.
+
+ my ( $match, # match to expand
+ $space, # search space
+ ) = @_;
+
+ # Returns nothing.
+
+ while ( $match->{ 'Q_END' } <= $space->{ 'Q_MAX' } and
+ $match->{ 'S_END' } <= $space->{ 'S_MAX' } and
+ uc substr( ${ $space->{ 'Q_SEQ' } }, $match->{ 'Q_END' }, 1 ) eq uc substr( ${ $space->{ 'S_SEQ' } }, $match->{ 'S_END' }, 1 ) )
+ {
+ $match->{ 'Q_END' }++;
+ $match->{ 'S_END' }++;
+ $match->{ 'LEN' }++;
}
- return wantarray ? @matches : \@matches;
+ $match->{ 'Q_END' }--;
+ $match->{ 'S_END' }--;
+ $match->{ 'LEN' }--;
}
-sub expand_match
+sub match_expand_backward
{
- # Martin A. Hansen, August 2006.
+ # Martin A. Hansen, August 2009.
- # Given two sequences and a match, expand the match maximally.
- # A match is defined like this: [ Q_BEG, Q_END, S_BEG, S_END ]
+ # Given a match and a search space hash,
+ # expand the match maximally in the backward
+ # direction.
- my ( $q_seq, # sequence 1 ref
- $s_seq, # sequence 2 ref
- $match, # sequence match
- $q_max, # q sequecne stop position
- $q_min, # q sequence start position
- $s_max, # s sequecne stop position
- $s_min, # s sequence start position
+ my ( $match, # match to expand
+ $space, # search space
) = @_;
- # returns match
+ # Returns nothing.
+
+ while ( $match->{ 'Q_BEG' } >= $space->{ 'Q_MIN' } and
+ $match->{ 'S_BEG' } >= $space->{ 'S_MIN' } and
+ uc substr( ${ $space->{ 'Q_SEQ' } }, $match->{ 'Q_BEG' }, 1 ) eq uc substr( ${ $space->{ 'S_SEQ' } }, $match->{ 'S_BEG' }, 1 ) )
+ {
+ $match->{ 'Q_BEG'}--;
+ $match->{ 'S_BEG'}--;
+ $match->{ 'LEN' }++;
+ }
+
+ $match->{ 'Q_BEG'}++;
+ $match->{ 'S_BEG'}++;
+ $match->{ 'LEN' }--;
+}
+
- my ( $q_pos, $s_pos );
+sub match_redundant
+{
+ # Martin A. Hansen, August 2009
+
+ my ( $new_match, # match
+ $redundant, # hashref
+ ) = @_;
- # expanding forward
+ # Returns boolean.
- $q_pos = $match->[ Q_END ] + 1;
- $s_pos = $match->[ S_END ] + 1;
+ my ( $match );
- while ( $q_pos <= $q_max and $s_pos <= $s_max and lc substr( ${ $q_seq }, $q_pos, 1 ) eq lc substr( ${ $s_seq }, $s_pos, 1 ) )
+ foreach $match ( @{ $redundant->{ $new_match->{ 'Q_BEG' } } } )
{
- $q_pos++;
- $s_pos++;
+ if ( $new_match->{ 'S_BEG' } >= $match->{ 'S_BEG' } and $new_match->{ 'S_END' } <= $match->{ 'S_END' } ) {
+ return 1;
+ }
}
- $match->[ Q_END ] = $q_pos - 1;
- $match->[ S_END ] = $s_pos - 1;
+ return 0;
+}
+
+
+sub match_redundant_add
+{
+ # Martin A. Hansen, August 2009
+
+ my ( $match, # match
+ $redundant, # hash ref
+ ) = @_;
+
+ # Returns nothing.
+
+ map { push @{ $redundant->{ $_ } }, $match } ( $match->{ 'Q_BEG' } .. $match->{ 'Q_END' } );
+}
+
+
+sub matches_filter
+{
+ # Martin A. Hansen, August 2009.
+
+ my ( $space, # search space
+ $matches, # list of matches
+ ) = @_;
+
+ # Returns a list.
+
+ my ( $best_match, $match, @new_matches );
- # expanding backwards
+ $best_match = shift @{ $matches };
- $q_pos = $match->[ Q_BEG ] - 1;
- $s_pos = $match->[ S_BEG ] - 1;
+ match_score( $best_match, $space );
- while ( $q_pos >= $q_min and $s_pos >= $s_min and lc substr( ${ $q_seq }, $q_pos, 1 ) eq lc substr( ${ $s_seq }, $s_pos, 1 ) )
+ foreach $match ( @{ $matches } )
{
- $q_pos--;
- $s_pos--;
+ match_score( $match, $space );
+
+ if ( $match->{ 'SCORE' } > 0 )
+ {
+ if ( $match->{ 'SCORE' } > $best_match->{ 'SCORE' } ) {
+ $best_match = $match;
+ } else {
+ push @new_matches, $match;
+ }
+ }
}
- $match->[ Q_BEG ] = $q_pos + 1;
- $match->[ S_BEG ] = $s_pos + 1;
+ unshift @new_matches, $best_match if $best_match->{ 'SCORE' } > 0;
- return $match;
+ return wantarray ? @new_matches : \@new_matches;
}
-sub score_match
+sub match_score
{
- # Martin A. Hansen, August 2006
+ # Martin A. Hansen, August 2009
- # given a match and a search space scores the match according to three criteria:
+ # Given a match and a search space scores the match according to three criteria:
# 1) length of match - the longer the better.
# 2) distance to closest corner - the shorter the better.
# 3) distance to closest narrow end of the search space - the shorter the better.
- # each of these scores are divided by search space dimentions, and the total score
- # is calculated: total = score_len - score_corner - score_narrow
-
- # the higher the score, the better the match.
+ # Each of these scores are divided by search space dimensions, and the total score
+ # is calculated: total = score_len - score_corner - score_narrow.
+ # The higher the score, the better the match.
+
my ( $match, # match
- $q_min, # q sequence start position
- $q_max, # q sequecne stop position
- $s_min, # s sequence start position
- $s_max, # s sequecne stop position
+ $space, # search space
) = @_;
-
- # returns a positive number
-
- my ( $q_dim, $s_dim, $score_len, $score_corner, $score_narrow, $score_total, $beg_diag_dist, $end_diag_dist,
- $min_diag_dist, $score_diag, $beg_narrow_dist, $end_narrow_dist, $max_narrow_dist );
- # ----- 1) scoring according to match length
+ # Returns nothing.
- $score_len = $match->[ LEN ] ** 3;
+ my ( $score_len, $score_corner, $score_diag, $score_narrow, $score_total );
- # ---- 2) score according to distance away from diagonal
+ $score_len = match_score_len( $match );
+ $score_corner = match_score_corner( $match, $space );
+ $score_diag = match_score_diag( $match, $space );
+ $score_narrow = match_score_narrow( $match, $space );
- $q_dim = $q_max - $q_min + 1;
- $s_dim = $s_max - $s_min + 1;
+ $score_total = $score_len - $score_corner - $score_diag - $score_narrow;
- if ( $q_dim >= $s_dim ) # s_dim is the narrow end
- {
- $beg_diag_dist = Maasha::Calc::dist_point2line( $match->[ Q_BEG ], $match->[ S_BEG ], $q_min, $s_min, $q_min + $s_dim, $s_min + $s_dim );
- $end_diag_dist = Maasha::Calc::dist_point2line( $match->[ Q_BEG ], $match->[ S_BEG ], $q_max - $s_dim, $s_max - $s_dim, $q_max, $s_max );
- }
- else
- {
- $beg_diag_dist = Maasha::Calc::dist_point2line( $match->[ Q_BEG ], $match->[ S_BEG ], $q_min, $s_min, $q_min + $q_dim, $s_min + $q_dim );
- $end_diag_dist = Maasha::Calc::dist_point2line( $match->[ Q_BEG ], $match->[ S_BEG ], $q_max - $q_dim, $s_max - $q_dim, $q_max, $s_max );
- }
+ printf STDERR "LEN: %d CORNER: %d DIAG: %d NARROW: %d TOTAL: %d\n", $score_len, $score_corner, $score_diag, $score_narrow, $score_total if VERBOSE;
- $min_diag_dist = Maasha::Calc::min( $beg_diag_dist, $end_diag_dist );
+ $match->{ 'SCORE' } = $score_total;
+}
- $score_diag = 2 * $min_diag_dist ** 2;
- # ----- 3) scoring according to distance to the narrow end of the search space
+sub match_score_len
+{
+ # Martin A. Hansen, August 2009
- if ( $q_dim > $s_dim ) # s_dim is the narrow end
- {
- $beg_narrow_dist = $match->[ Q_BEG ] - $q_min;
- $end_narrow_dist = $q_max - $match->[ Q_BEG ];
+ # Score according to match length.
- $max_narrow_dist = Maasha::Calc::max( $beg_narrow_dist, $end_narrow_dist );
- }
- elsif ( $q_dim < $s_dim )
- {
- $beg_narrow_dist = $match->[ S_BEG ] - $s_min;
- $end_narrow_dist = $s_max - $match->[ S_BEG ];
+ my ( $match, # match
+ ) = @_;
- $max_narrow_dist = Maasha::Calc::max( $beg_narrow_dist, $end_narrow_dist );
- }
- else
- {
- $max_narrow_dist = 0;
- }
+ # Returns integer.
- $score_narrow = $max_narrow_dist;
+ return $match->{ 'LEN' } * SCORE_FACTOR_LEN;
+}
+
+
+sub match_score_corner
+{
+ # Martin A. Hansen, August 2009
+
+ # Score according to distance from corner.
- $score_total = $score_len - $score_narrow - $score_diag;
+ my ( $match, # match
+ $space, # search space
+ ) = @_;
+
+ # Returns integer.
+
+ my ( $left_dist, $right_dist, $score_corner );
+
+ $left_dist = Maasha::Calc::dist_point2point( $space->{ 'Q_MIN' }, $space->{ 'S_MIN' }, $match->{ 'Q_BEG' } , $match->{ 'S_BEG' } );
+ $right_dist = Maasha::Calc::dist_point2point( $space->{ 'Q_MAX' }, $space->{ 'S_MAX' }, $match->{ 'Q_END' } , $match->{ 'S_END' } );
- return $score_total;
+ $score_corner = Maasha::Calc::min( $left_dist, $right_dist );
+
+ return $score_corner * SCORE_FACTOR_CORNER;
}
-sub score_match_niels
+sub match_score_diag
{
- # Niels Larsen, June 2004.
-
- # Creates a crude "heuristic" attempt of telling how likely it is that a
- # given match occurs by chance in a given search space. If sequences are
- # given their composition is taken into account. The scoring punishes
- # distance from diagonal(s) and distance from previous match(es). Scores
- # range from zero and up, and lower is better.
-
- my ( $match, # Match array
- $q_seq, # Either q_seq or s_seq
- $q_min, # Lower bound search area (query sequence)
- $q_max, # Upper bound search area (query sequence)
- $s_min, # Lower bound search area (subject sequence)
- $s_max, # Upper bound search area (subject sequence)
- ) = @_;
-
- # Returns a positive number.
-
- my ( $q_beg, $s_beg, $q_end, $s_end, $q_dim, $s_dim, $seq, $pos,
- $q_delta_beg, $s_delta_beg, $q_delta_end, $s_delta_end, $i,
- $delta_beg_max, $delta_end_max, $as, $gs, $ts, $cs, $pmatch,
- $score, $moves, $dist_beg, $dist_end, $seqlen, %chars, $delta,
- $delta_max );
-
- $q_beg = $match->[Q_BEG];
- $q_end = $match->[Q_END];
- $s_beg = $match->[S_BEG];
- $s_end = $match->[S_END];
-
- # >>>>>>>>>>>>>>>>>>>>>>> CRUDE INITIAL SCORE <<<<<<<<<<<<<<<<<<<<<<
+ # Martin A. Hansen, August 2009
- # Get the probability of a match from the sequence composition (when
- # match is longer than 20 and sequence is given, otherwise use 0.25)
- # and raise that to the power of the length.
+ # Score according to distance from diagonal.
- if ( $match->[LEN] > 20 and defined $q_seq )
- {
- $seq = substr ${ $q_seq }, $q_beg, $q_end-$q_beg+1;
- $seqlen = length $seq;
+ my ( $match, # match
+ $space, # search space
+ ) = @_;
- $chars{"a"} = $chars{"g"} = $chars{"c"} = $chars{"t"} = 0;
+ # Returns integer.
- for ( $i = 0; $i < $seqlen; $i++ )
- {
- $chars{ substr $seq, $i, 1 }++;
- }
+ my ( $q_dim, $s_dim, $beg_diag_dist, $end_diag_dist, $min_diag_dist, $score_diag );
+
+ $q_dim = $space->{ 'Q_MAX' } - $space->{ 'Q_MIN' } + 1;
+ $s_dim = $space->{ 'S_MAX' } - $space->{ 'S_MIN' } + 1;
- $pmatch = ($chars{"a"}/$seqlen)**2 + ($chars{"c"}/$seqlen)**2
- + ($chars{"g"}/$seqlen)**2 + ($chars{"t"}/$seqlen)**2;
+ if ( $q_dim >= $s_dim ) # s_dim is the narrow end
+ {
+ $beg_diag_dist = Maasha::Calc::dist_point2line( $match->{ 'Q_BEG' },
+ $match->{ 'S_BEG' },
+ $space->{ 'Q_MIN' },
+ $space->{ 'S_MIN' },
+ $space->{ 'Q_MIN' } + $s_dim,
+ $space->{ 'S_MIN' } + $s_dim );
+
+ $end_diag_dist = Maasha::Calc::dist_point2line( $match->{ 'Q_BEG' },
+ $match->{ 'S_BEG' },
+ $space->{ 'Q_MAX' } - $s_dim,
+ $space->{ 'S_MAX' } - $s_dim,
+ $space->{ 'Q_MAX' },
+ $space->{ 'S_MAX' } );
}
- else {
- $pmatch = 0.25;
+ else
+ {
+ $beg_diag_dist = Maasha::Calc::dist_point2line( $match->{ 'Q_BEG' },
+ $match->{ 'S_BEG' },
+ $space->{ 'Q_MIN' },
+ $space->{ 'S_MIN' },
+ $space->{ 'Q_MIN' } + $q_dim,
+ $space->{ 'S_MIN' } + $q_dim );
+
+ $end_diag_dist = Maasha::Calc::dist_point2line( $match->{ 'Q_BEG' },
+ $match->{ 'S_BEG' },
+ $space->{ 'Q_MAX' } - $q_dim,
+ $space->{ 'S_MAX' } - $q_dim,
+ $space->{ 'Q_MAX' },
+ $space->{ 'S_MAX' } );
}
- $score = $pmatch ** ( $q_end - $q_beg + 1 );
+ $min_diag_dist = Maasha::Calc::min( $beg_diag_dist, $end_diag_dist );
-# # Punish by difference in height and width of search space,
-
-# $q_dim = $q_max - $q_min + 1;
-# $s_dim = $s_max - $s_min + 1;
-
-# if ( $q_dim != $s_dim ) {
-# $score *= abs ( $q_dim - $s_dim ) ** 2;
-# }
-
- return $score if $score > 0.25;
+ $score_diag = $min_diag_dist;
- # Punish by how far the match is to the closest corner of the search
- # space,
+ return $score_diag * SCORE_FACTOR_DIAG;
+}
- $q_delta_beg = $q_beg - $q_min;
- $s_delta_beg = $s_beg - $s_min;
-
- $q_delta_end = $q_max - $q_end;
- $s_delta_end = $s_max - $s_end;
-
- if ( $q_delta_beg > $s_delta_beg ) {
- $delta_beg_max = $q_delta_beg;
- } else {
- $delta_beg_max = $s_delta_beg;
- }
- if ( $q_delta_end > $s_delta_end ) {
- $delta_end_max = $q_delta_end;
- } else {
- $delta_end_max = $s_delta_end;
- }
+sub match_score_narrow
+{
+ # Martin A. Hansen, August 2009
- if ( $delta_beg_max <= $delta_end_max ) {
- $score *= ($delta_beg_max+1) ** 2.0;
- } else {
- $score *= ($delta_end_max+1) ** 2.0;
- }
+ # Score according to distance to the narrow end of the search space.
+
+ my ( $match, # match
+ $space, # search space
+ ) = @_;
+
+ # Returns integer.
+
+ my ( $max_narrow_dist, $q_dim, $s_dim, $beg_narrow_dist, $end_narrow_dist, $score_narrow );
- return $score if $score > 0.25;
+ $max_narrow_dist = 0;
- # Add penalty if the match is towards the narrow end of the
- # search space,
+ $q_dim = $space->{ 'Q_MAX' } - $space->{ 'Q_MIN' } + 1;
+ $s_dim = $space->{ 'S_MAX' } - $space->{ 'S_MIN' } + 1;
- if ( ($q_max - $q_min) <= ($s_max - $s_min) )
+ if ( $q_dim > $s_dim ) # s_dim is the narrow end
{
- if ( $q_delta_beg > $s_delta_beg )
- {
- $score *= 2 * ( $q_delta_beg - $s_delta_beg ) ** 3;
- }
- elsif ( $q_delta_end > $s_delta_end )
- {
- $score *= 2 * ( $q_delta_end - $s_delta_end ) ** 3;
- }
+ $beg_narrow_dist = $match->{ 'Q_BEG' } - $space->{ 'Q_MIN' };
+ $end_narrow_dist = $space->{ 'Q_MAX' } - $match->{ 'Q_BEG' };
+
+ $max_narrow_dist = Maasha::Calc::max( $beg_narrow_dist, $end_narrow_dist );
}
- else
+ elsif ( $q_dim < $s_dim ) # q_dim is the narrow end
{
- if ( $s_delta_beg > $q_delta_beg )
- {
- $score *= 2 * ( $s_delta_beg - $q_delta_beg ) ** 3;
- }
- elsif ( $s_delta_end > $q_delta_end )
- {
- $score *= 2 * ( $s_delta_end - $q_delta_end ) ** 3;
- }
- }
+ $beg_narrow_dist = $match->{ 'S_BEG' } - $space->{ 'S_MIN' };
+ $end_narrow_dist = $space->{ 'S_MAX' } - $match->{ 'S_BEG' };
- if ( $score < 0 ) {
- print STDERR "q_min, q_max, s_min, s_max: $q_min, $q_max, $s_min, $s_max\n";
- die qq (Score <= 0 -> $score);
+ $max_narrow_dist = Maasha::Calc::max( $beg_narrow_dist, $end_narrow_dist );
}
- return $score;
+ $score_narrow = $max_narrow_dist;
+
+ return $score_narrow * SCORE_FACTOR_NARROW;
}
-sub insert_indels
+sub shift_case
{
- # Martin A. Hansen, June 2009.
+ # Martin A. Hansen, August 2009.
- # Given a list of matches and references to q_seq and s_seq,
- # insert indels to form aligned sequences.
+ # Given a list of matches and sequence references
+ # uppercase only matching sequnce while lowercasing the rest.
my ( $matches, # list of matches
- $q_seq, # ref to query sequence
- $s_seq, # ref to subject sequence
+ $q_seq, # query sequence ref
+ $s_seq, # subject sequence ref
) = @_;
# Returns nothing.
- my ( $q_indels, $s_indels, $match, $diff, $first );
-
- @{ $matches } = sort { $a->[ Q_BEG ] <=> $b->[ Q_BEG ] } @{ $matches };
+ my ( $match );
- $first = 1;
- $q_indels = 0;
- $s_indels = 0;
+ ${ $q_seq } = lc ${ $q_seq };
+ ${ $s_seq } = lc ${ $s_seq };
- # ---- initial indels ----
+ foreach $match ( @{ $matches } )
+ {
+ substr ${ $q_seq }, $match->{ 'Q_BEG' }, $match->{ 'LEN' }, uc substr ${ $q_seq }, $match->{ 'Q_BEG' }, $match->{ 'LEN' };
+ substr ${ $s_seq }, $match->{ 'S_BEG' }, $match->{ 'LEN' }, uc substr ${ $s_seq }, $match->{ 'S_BEG' }, $match->{ 'LEN' };
+ }
+}
- $match = shift @{ $matches };
- substr ${ $q_seq }, $match->[ Q_BEG ] + $q_indels, $match->[ LEN ], uc substr ${ $q_seq }, $match->[ Q_BEG ] + $q_indels, $match->[ LEN ];
- substr ${ $s_seq }, $match->[ S_BEG ] + $s_indels, $match->[ LEN ], uc substr ${ $s_seq }, $match->[ S_BEG ] + $s_indels, $match->[ LEN ];
- $diff = ( $match->[ Q_BEG ] + $q_indels ) - ( $match->[ S_BEG ] + $s_indels );
+sub insert_indels
+{
+ # Martin A. Hansen, August 2009.
- if ( $diff < 0 )
- {
- substr ${ $q_seq }, 0, 0, '-' x abs $diff;
+ # Given a list of matches and sequence references
+ # insert indels to form aligned sequences.
- $q_indels += abs $diff;
- }
- elsif ( $diff > 0 )
- {
- substr ${ $s_seq }, 0, 0, '-' x abs $diff;
+ my ( $matches, # list of matches
+ $q_seq, # query sequence ref
+ $s_seq, # subject sequence ref
+ ) = @_;
- $s_indels += abs $diff;
- }
+ my ( $i, $q_dist, $s_dist, $q_indels, $s_indels, $diff );
- # ---- internal indels ----
+ $q_indels = 0;
+ $s_indels = 0;
- foreach $match ( @{ $matches } )
+ if ( @{ $matches } > 0 )
{
- substr ${ $q_seq }, $match->[ Q_BEG ] + $q_indels, $match->[ LEN ], uc substr ${ $q_seq }, $match->[ Q_BEG ] + $q_indels, $match->[ LEN ];
- substr ${ $s_seq }, $match->[ S_BEG ] + $s_indels, $match->[ LEN ], uc substr ${ $s_seq }, $match->[ S_BEG ] + $s_indels, $match->[ LEN ];
+ @{ $matches } = sort { $a->{ 'Q_BEG' } <=> $b->{ 'Q_BEG' } } @{ $matches }; # FIXME is this necessary?
+
+ # >>>>>>>>>>>>>> FIRST MATCH <<<<<<<<<<<<<<
- $diff = ( $match->[ Q_BEG ] + $q_indels ) - ( $match->[ S_BEG ] + $s_indels );
+ $diff = $matches->[ 0 ]->{ 'Q_BEG' } - $matches->[ 0 ]->{ 'S_BEG' };
+
+ if ( $diff > 0 )
+ {
+ substr ${ $s_seq }, 0, 0, '-' x $diff;
- if ( $diff < 0 )
+ $s_indels += $diff;
+ }
+ elsif ( $diff < 0 )
{
- substr ${ $q_seq }, $match->[ Q_BEG ] + $q_indels, 0, '-' x abs $diff;
+ substr ${ $q_seq }, 0, 0, '-' x abs $diff;
$q_indels += abs $diff;
}
- elsif ( $diff > 0 )
+
+ # >>>>>>>>>>>>>> MIDDLE MATCHES <<<<<<<<<<<<<<
+
+ for ( $i = 0; $i < scalar @{ $matches } - 1; $i++ )
{
- substr ${ $s_seq }, $match->[ S_BEG ] + $s_indels, 0, '-' x abs $diff;
+ $q_dist = $matches->[ $i + 1 ]->{ 'Q_BEG' } - $matches->[ $i ]->{ 'Q_END' };
+ $s_dist = $matches->[ $i + 1 ]->{ 'S_BEG' } - $matches->[ $i ]->{ 'S_END' };
+
+ $diff = $q_dist - $s_dist;
+
+ if ( $diff > 0 )
+ {
+ substr ${ $s_seq }, $s_indels + $matches->[ $i ]->{ 'S_END' } + 1, 0, '-' x $diff;
- $s_indels += abs $diff;
+ $s_indels += $diff;
+ }
+ elsif ( $diff < 0 )
+ {
+ substr ${ $q_seq }, $q_indels + $matches->[ $i ]->{ 'Q_END' } + 1, 0, '-' x abs $diff;
+
+ $q_indels += abs $diff;
+ }
}
}
- # ---- last indels ----
+ # >>>>>>>>>>>>>> LAST MATCH <<<<<<<<<<<<<<
- $diff = length( ${ $s_seq } ) - length( ${ $q_seq } );
+ $diff = length( ${ $q_seq } ) - length( ${ $s_seq } );
- if ( $diff > 0 ) {
- ${ $q_seq } .= '-' x abs $diff;
- } else {
- ${ $s_seq } .= '-' x abs $diff;
+ if ( $diff > 0 )
+ {
+ ${ $s_seq } .= '-' x $diff;
+
+ $s_indels += $diff;
+ }
+ else
+ {
+ ${ $q_seq } .= '-' x abs $diff;
+
+ $q_indels += abs $diff;
}
}
+1;
+
+__END__
+
sub align_sim_local
{
}
-sub align_consensus
-{
- # Martin A. Hansen, June 2006.
-
- # Given an alignment as a list of FASTA entries,
- # generates a consensus sequences based on the
- # entropies for each column similar to the way
- # a sequence logo i calculated. Returns the
- # consensus sequence as a FASTA entry.
-
- my ( $entries, # list of aligned FASTA entries
- $type, # residue type - OPTIONAL
- $min_sim, # minimum similarity - OPTIONAL
- ) = @_;
-
- # Returns tuple
-
- my ( $bit_max, $data, $pos, $char, $score, $entry );
-
- $type ||= Maasha::Seq::seq_guess_type( $entries->[ 0 ] );
- $min_sim ||= 50;
-
- if ( $type =~ /protein/ ) {
- $bit_max = 4;
- } else {
- $bit_max = 2;
- }
-
- $data = Maasha::Seq::seqlogo_calc( $bit_max, $entries );
-
- foreach $pos ( @{ $data } )
- {
- ( $char, $score ) = @{ $pos->[ -1 ] };
-
- if ( ( $score / $bit_max ) * 100 >= $min_sim ) {
- $entry->[ SEQ ] .= $char;
- } else {
- $entry->[ SEQ ] .= "-";
- }
- }
-
- $entry->[ HEAD ] = "Consensus: $min_sim%";
-
- return wantarray ? @{ $entry } : $entry;
-}
-
-
# >>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>>><<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<<