# $Id: GenericHSP.pm 16123 2009-09-17 12:57:27Z cjfields $
#
# BioPerl module for Bio::Search::HSP::GenericHSP
#
# Please direct questions and support issues to <bioperl-l@bioperl.org> 
#
# Cared for by Jason Stajich <jason@bioperl.org>
#
# 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::Search::HSP::GenericHSP - A "Generic" implementation of a High Scoring Pair

=head1 SYNOPSIS

    my $hsp = Bio::Search::HSP::GenericHSP->new( -algorithm => 'blastp',
                                                -evalue    => '1e-30',
                                                );

    $r_type = $hsp->algorithm;

    $pvalue = $hsp->p();

    $evalue = $hsp->evalue();

    $frac_id = $hsp->frac_identical( ['query'|'hit'|'total'] );

    $frac_cons = $hsp->frac_conserved( ['query'|'hit'|'total'] );

    $gaps = $hsp->gaps( ['query'|'hit'|'total'] );

    $qseq = $hsp->query_string;

    $hseq = $hsp->hit_string;

    $homo_string = $hsp->homology_string;

    $len = $hsp->length( ['query'|'hit'|'total'] );

    $len = $hsp->length( ['query'|'hit'|'total'] );

    $rank = $hsp->rank;

# TODO: Describe how to configure a SearchIO stream so that it generates
#       GenericHSP objects.

=head1 DESCRIPTION

This implementation is "Generic", meaning it is is suitable for
holding information about High Scoring pairs from most Search reports
such as BLAST and FastA.  Specialized objects can be derived from
this.

Unless you're writing a parser, you won't ever need to create a
GenericHSP or any other HSPI-implementing object. If you use
the SearchIO system, HSPI objects are created automatically from
a SearchIO stream which returns Bio::Search::Result::ResultI objects
and you get the HSPI objects via the ResultI API.

For documentation on what you can do with GenericHSP (and other HSPI
objects), please see the API documentation in
L<Bio::Search::HSP::HSPI|Bio::Search::HSP::HSPI>.

=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:

  http://bugzilla.open-bio.org/

=head1 AUTHOR - Jason Stajich and Steve Chervitz

Email jason-at-bioperl.org
Email sac-at-bioperl.org

=head1 CONTRIBUTORS

Sendu Bala, bix@sendu.me.uk

=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::Search::HSP::GenericHSP;
use strict;

use Bio::Root::Root;
use Bio::SeqFeature::Similarity;

use base qw(Bio::Search::HSP::HSPI);

=head2 new

 Title   : new
 Usage   : my $obj = Bio::Search::HSP::GenericHSP->new();
 Function: Builds a new Bio::Search::HSP::GenericHSP object
 Returns : Bio::Search::HSP::GenericHSP
 Args    : -algorithm => algorithm used (BLASTP, TBLASTX, FASTX, etc)
           -evalue    => evalue
           -pvalue    => pvalue
           -bits      => bit value for HSP
           -score     => score value for HSP (typically z-score but depends on
                                              analysis)
           -hsp_length=> Length of the HSP (including gaps)
           -identical => # of residues that that matched identically
           -percent_identity => (optional) percent identity
           -conserved => # of residues that matched conservatively
                           (only protein comparisions;
                            conserved == identical in nucleotide comparisons)
           -hsp_gaps   => # of gaps in the HSP
           -query_gaps => # of gaps in the query in the alignment
           -hit_gaps   => # of gaps in the subject in the alignment
           -query_name  => HSP Query sequence name (if available)
           -query_start => HSP Query start (in original query sequence coords)
           -query_end   => HSP Query end (in original query sequence coords)
           -query_length=> total length of the query sequence
           -query_seq   => query sequence portion of the HSP
           -query_desc  => textual description of the query
           -hit_name    => HSP Hit sequence name (if available)
           -hit_start   => HSP Hit start (in original hit sequence coords)
           -hit_end     => HSP Hit end (in original hit sequence coords)
           -hit_length  => total length of the hit sequence
           -hit_seq     => hit sequence portion of the HSP
           -hit_desc    => textual description of the hit
           -homology_seq=> homology sequence for the HSP
           -hit_frame   => hit frame (only if hit is translated protein)
           -query_frame => query frame (only if query is translated protein)
           -rank        => HSP rank
           -links       => HSP links information (WU-BLAST only)
           -hsp_group   => HSP Group informat (WU-BLAST only)
           -gap_symbol  => symbol representing a gap (default = '-')
           -hit_features=> string of features found in or near HSP hit
                           region (reported in some BLAST text output,
                           v. 2.2.13 and up)
           -stranded    => If the algorithm isn't known (i.e. defaults to
                           'generic'), setting this will indicate start/end
                           coordinates are to be used to determine the strand
                           for 'query', 'subject', 'hit', 'both', or 'none'
                           (default = 'none')

=cut

sub new {
    my($class,%args) = @_;

    # don't pass anything to SUPER; complex hierarchy results in lots of work
    # for nothing
    
    my $self = $class->SUPER::new();

    # for speed, don't use _rearrange and just store all input data directly
    # with no method calls and no work done. work can be carried
    # out just-in-time later if desired
    while (my ($arg, $value) = each %args) {
        $arg =~ tr/a-z\055/A-Z/d;
        $self->{$arg} = $value;
    }
    my $bits = $self->{BITS};

    defined $self->{VERBOSE} && $self->verbose($self->{VERBOSE});
    if (exists $self->{GAP_SYMBOL}) {
        # not checking anything else but the length (must be 1 as only one gap
        # symbol allowed currently); can add in support for symbol checks or
        # multiple symbols later if needed
        $self->throw("Gap symbol must be of length 1") if
            CORE::length($self->{GAP_SYMBOL}) != 1;
    } else {
        # dafault
        $self->{GAP_SYMBOL} = '-';
    }
    $self->{ALGORITHM} ||= 'GENERIC';
    $self->{STRANDED} ||= 'NONE';
    
    if (! defined $self->{QUERY_LENGTH} || ! defined $self->{HIT_LENGTH}) {
        $self->throw("Must define hit and query length");
    }

    $self->{'_sequenceschanged'} = 1;
    
    $self->{_finished_new} = 1;
    return $self;
}

sub _logical_length {
    my ($self, $type) = @_;
    my $algo = $self->algorithm;
    my $len = $self->length($type);
    my $logical = $len;
    if($algo =~ /^(PSI)?T(BLAST|FAST)[NY]/oi ) {
        $logical = $len/3 if $type =~ /sbjct|hit|tot/i;
    } elsif($algo =~ /^(BLAST|FAST)(X|Y|XY)/oi ) {
        $logical = $len/3 if $type =~ /query|tot/i;
    } elsif($algo =~ /^T(BLAST|FAST)(X|Y|XY)/oi ) {
        $logical = $len/3;
    }
    return $logical;
}

=head2 L<Bio::Search::HSP::HSPI> methods

Implementation of L<Bio::Search::HSP::HSPI> methods follow

=head2 algorithm

 Title   : algorithm
 Usage   : my $r_type = $hsp->algorithm
 Function: Obtain the name of the algorithm used to obtain the HSP
 Returns : string (e.g., BLASTP)
 Args    : [optional] scalar string to set value

=cut

sub algorithm{
    my ($self,$value) = @_;
    my $previous = $self->{'ALGORITHM'};
    if( defined $value || ! defined $previous ) {
        $value = $previous = '' unless defined $value;
        $self->{'ALGORITHM'} = $value;
    }

    return $previous;
}

=head2 pvalue

 Title   : pvalue
 Usage   : my $pvalue = $hsp->pvalue();
 Function: Returns the P-value for this HSP or undef
 Returns : float or exponential (2e-10)
           P-value is not defined with NCBI Blast2 reports.
 Args    : [optional] numeric to set value

=cut

sub pvalue {
    my ($self,$value) = @_;
    my $previous = $self->{'PVALUE'};
    if( defined $value  ) {
        $self->{'PVALUE'} = $value;
    }
    return $previous;
}

=head2 evalue

 Title   : evalue
 Usage   : my $evalue = $hsp->evalue();
 Function: Returns the e-value for this HSP
 Returns : float or exponential (2e-10)
 Args    : [optional] numeric to set value

=cut

sub evalue {
    my ($self,$value) = @_;
    my $previous = $self->{'EVALUE'};
    if( defined $value  ) {
        $self->{'EVALUE'} = $value;
    }
    return $previous;
}

=head2 frac_identical

 Title   : frac_identical
 Usage   : my $frac_id = $hsp->frac_identical( ['query'|'hit'|'total'] );
 Function: Returns the fraction of identitical positions for this HSP
 Returns : Float in range 0.0 -> 1.0
 Args    : arg 1:  'query' = num identical / length of query seq (without gaps)
                   'hit'   = num identical / length of hit seq (without gaps)
                             synonyms: 'sbjct', 'subject'
                   'total' = num identical / length of alignment (with gaps)
                             synonyms: 'hsp'
                   default = 'total'
           arg 2: [optional] frac identical value to set for the type requested

=cut

sub frac_identical {
   my ($self, $type,$value) = @_;

    unless ($self->{_did_prefrac}) {
        $self->_pre_frac;
    }

   $type = lc $type if defined $type;
   $type = 'hit' if( defined $type &&
		     $type =~ /subject|sbjct/);
   $type = 'total' if( ! defined $type || $type eq 'hsp' ||
                        $type !~ /query|hit|subject|sbjct|total/);
   my $previous = $self->{'_frac_identical'}->{$type};
   if( defined $value || ! defined $previous ) {
       $value = $previous = '' unless defined $value;
       if( $type eq 'hit' || $type eq 'query' ) {
           $self->$type()->frac_identical( $value);
       }
       $self->{'_frac_identical'}->{$type} = $value;
   }
   return $previous;

}

=head2 frac_conserved

 Title    : frac_conserved
 Usage    : my $frac_cons = $hsp->frac_conserved( ['query'|'hit'|'total'] );
 Function : Returns the fraction of conserved positions for this HSP.
            This is the fraction of symbols in the alignment with a
            positive score.
 Returns : Float in range 0.0 -> 1.0
 Args    : arg 1: 'query' = num conserved / length of query seq (without gaps)
                  'hit'   = num conserved / length of hit seq (without gaps)
                             synonyms: 'sbjct', 'subject'
                  'total' = num conserved / length of alignment (with gaps)
                             synonyms: 'hsp'
                  default = 'total'
           arg 2: [optional] frac conserved value to set for the type requested

=cut

sub frac_conserved {
    my ($self, $type,$value) = @_;

    unless ($self->{_did_prefrac}) {
        $self->_pre_frac;
    }

    $type = lc $type if defined $type;
    $type = 'hit' if( defined $type && $type =~ /subject|sbjct/);
    $type = 'total' if( ! defined $type || $type eq 'hsp' ||
                        $type !~ /query|hit|subject|sbjct|total/);
    my $previous = $self->{'_frac_conserved'}->{$type};
    if( defined $value || ! defined $previous ) {
        $value = $previous = '' unless defined $value;
        $self->{'_frac_conserved'}->{$type} = $value;
    }
    return $previous;
}

=head2 gaps

 Title    : gaps
 Usage    : my $gaps = $hsp->gaps( ['query'|'hit'|'total'] );
 Function : Get the number of gap characters in the query, hit, or total alignment.
 Returns  : Integer, number of gaps or 0 if none
 Args     : arg 1: 'query' = num gap characters in query seq
                   'hit'   = num gap characters in hit seq; synonyms: 'sbjct', 'subject'
                   'total' = num gap characters in whole alignment;  synonyms: 'hsp'
                   default = 'total'
            arg 2: [optional] integer gap value to set for the type requested

=cut

sub gaps {
    my ($self, $type, $value) = @_;

    unless ($self->{_did_pregaps}) {
        $self->_pre_gaps;
    }

    $type = lc $type if defined $type;
    $type = 'total' if( ! defined $type || $type eq 'hsp' ||
                        $type !~ /query|hit|subject|sbjct|total/);
    $type = 'hit' if $type =~ /sbjct|subject/;
    my $previous = $self->{'_gaps'}->{$type};
    if( defined $value || ! defined $previous ) {
        $value = $previous = '' unless defined $value;
        $self->{'_gaps'}->{$type} = $value;
    }
    return $previous || 0;
}

=head2 query_string

 Title   : query_string
 Usage   : my $qseq = $hsp->query_string;
 Function: Retrieves the query sequence of this HSP as a string
 Returns : string
 Args    : [optional] string to set for query sequence


=cut

sub query_string{
    my ($self,$value) = @_;
    my $previous = $self->{QUERY_SEQ};
    if( defined $value || ! defined $previous ) {
        $value = $previous = '' unless defined $value;
        $self->{QUERY_SEQ} = $value;
        # do some housekeeping so we know when to
        # re-run _calculate_seq_positions
        $self->{'_sequenceschanged'} = 1;
    }
    return $previous;
}

=head2 hit_string

 Title   : hit_string
 Usage   : my $hseq = $hsp->hit_string;
 Function: Retrieves the hit sequence of this HSP as a string
 Returns : string
 Args    : [optional] string to set for hit sequence


=cut

sub hit_string{
    my ($self,$value) = @_;
    my $previous = $self->{HIT_SEQ};
    if( defined $value || ! defined $previous ) {
        $value = $previous = '' unless defined $value;
        $self->{HIT_SEQ} = $value;
        # do some housekeeping so we know when to
        # re-run _calculate_seq_positions
        $self->{'_sequenceschanged'} = 1;
    }
    return $previous;
}

=head2 homology_string

 Title   : homology_string
 Usage   : my $homo_string = $hsp->homology_string;
 Function: Retrieves the homology sequence for this HSP as a string.
         : The homology sequence is the string of symbols in between the
         : query and hit sequences in the alignment indicating the degree
         : of conservation (e.g., identical, similar, not similar).
 Returns : string
 Args    : [optional] string to set for homology sequence

=cut

sub homology_string{
    my ($self,$value) = @_;
    my $previous = $self->{HOMOLOGY_SEQ};
    if( defined $value || ! defined $previous ) {
        $value = $previous = '' unless defined $value;
        $self->{HOMOLOGY_SEQ} = $value;
        # do some housekeeping so we know when to
        # re-run _calculate_seq_positions
        $self->{'_sequenceschanged'} = 1;
    }
    return $previous;
}

=head2 length

 Title    : length
 Usage    : my $len = $hsp->length( ['query'|'hit'|'total'] );
 Function : Returns the length of the query or hit in the alignment
            (without gaps)
            or the aggregate length of the HSP (including gaps;
            this may be greater than either hit or query )
 Returns  : integer
 Args     : arg 1: 'query' = length of query seq (without gaps)
                   'hit'   = length of hit seq (without gaps) (synonyms: sbjct, subject)
                   'total' = length of alignment (with gaps)
                   default = 'total'
            arg 2: [optional] integer length value to set for specific type

=cut

sub length {

    my $self = shift;
    my $type = shift;

    $type = 'total' unless defined $type;
    $type = lc $type;

    if( $type =~ /^q/i ) {
        return $self->query()->length(shift);
    } elsif( $type =~ /^(hit|subject|sbjct)/ ) {
        return $self->hit()->length(shift);
    } else {
        my $v = shift;
        if( defined $v ) {
            $self->{HSP_LENGTH} = $v;
        }
        return $self->{HSP_LENGTH};
   }
    return 0; # should never get here
}

=head2 hsp_length

 Title   : hsp_length
 Usage   : my $len = $hsp->hsp_length()
 Function: shortcut  length('hsp')
 Returns : floating point between 0 and 100
 Args    : none

=cut

sub hsp_length { return shift->length('hsp', shift); }

=head2 percent_identity

 Title   : percent_identity
 Usage   : my $percentid = $hsp->percent_identity()
 Function: Returns the calculated percent identity for an HSP
 Returns : floating point between 0 and 100
 Args    : none


=cut

sub percent_identity {
    my $self = shift;

    unless ($self->{_did_prepi}) {
        $self->_pre_pi;
    }

    return $self->SUPER::percent_identity(@_);
}

=head2 frame

 Title   : frame
 Usage   : my ($qframe, $hframe) = $hsp->frame('list',$queryframe,$subjectframe)
 Function: Set the Frame for both query and subject and insure that
           they agree.
           This overrides the frame() method implementation in
           FeaturePair.
 Returns : array of query and subject frame if return type wants an array
           or query frame if defined or subject frame if not defined
 Args    : 'hit' or 'subject' or 'sbjct' to retrieve the frame of the subject (default)
           'query' to retrieve the query frame 
           'list' or 'array' to retrieve both query and hit frames together
 Note    : Frames are stored in the GFF way (0-2) not 1-3
           as they are in BLAST (negative frames are deduced by checking
                                 the strand of the query or hit)

=cut

# Note: changed 4/19/08 - bug 2485
#
# frame() is supposed to be a getter/setter (as is implied by the Function desc
# above; this is also consistent with Bio::SeqFeature::SimilarityPair).  Also,
# the API is not consistent with the other HSP/SimilarityPair methods such as
# strand(), start(), end(), etc. 
#
# In order to make this consistent with other methods all work is now done
# when the features are instantiated and not delayed.  We compromise by
# defaulting back 'to hit' w/o passed args.  Setting is now allowed.  

sub frame {
    my $self = shift;
    my $val = shift;
    if (!defined $val) {
        # unfortunately, w/o args we need to warn about API changes
        $self->warn("API for frame() has changed.\n".
                    "Please refer to documentation for Bio::Search::HSP::GenericHSP;\n".
                    "returning query frame");
        $val = 'query';
    }
    $val =~ s/^\s+//;

    if( $val =~ /^q/i ) {
        return $self->query->frame(@_);
    } elsif( $val =~ /^hi|^s/i ) {
        return $self->hit->frame(@_);
    } elsif (  $val =~ /^list|array/i ) {
        return ($self->query->frame($_[0]), 
            $self->hit->frame($_[1]) );
    } elsif ( $val =~ /^\d+$/) {
        # old API i.e. frame($query_frame, $hit_frame). This catches all but one
        # case, where no arg is passed (so the hit is wanted).  
        $self->warn("API for frame() has changed.\n".
                    "Please refer to documentation for Bio::Search::HSP::GenericHSP");
        wantarray ? 
        return ($self->query->frame($val), 
            $self->hit->frame(@_) ) :
        return $self->hit->frame($val,@_);
    } else { 
        $self->warn("unrecognized component '$val' requested\n");
    }
    return 0;
}

=head2 get_aln

 Title   : get_aln
 Usage   : my $aln = $hsp->gel_aln
 Function: Returns a L<Bio::SimpleAlign> object representing the HSP alignment
 Returns : L<Bio::SimpleAlign>
 Args    : none

=cut

sub get_aln {
    my ($self) = @_;
    require Bio::LocatableSeq;
    require Bio::SimpleAlign;
    my $aln = Bio::SimpleAlign->new();
    my $hs = $self->hit_string();
    my $qs = $self->query_string();
    # FASTA specific stuff moved to the FastaHSP object
    my $seqonly = $qs;
    $seqonly =~ s/[\-\s]//g;
    my ($q_nm,$s_nm) = ($self->query->seq_id(),
                        $self->hit->seq_id());
    # Should we silently change the name of the query or hit if it isn't
    # defined?  May need revisiting... cjfields 2008-12-3 (commented out below)
    
    #unless( defined $q_nm && CORE::length ($q_nm) ) {
    #    $q_nm = 'query';
    #}
    #unless( defined $s_nm && CORE::length ($s_nm) ) {
    #    $s_nm = 'hit';
    #}
    
    # mapping: 1 residues for every x coordinate positions
    my $query = Bio::LocatableSeq->new('-seq'   => $qs,
                                      '-id'    => $q_nm,
                                      '-start' => $self->query->start,
                                      '-end'   => $self->query->end,
                                      '-force_nse' => $q_nm ? 0 : 1,
                                      '-mapping' => [1, $self->{_query_mapping}]
                                      );
    $seqonly = $hs;
    $seqonly =~ s/[\-\s]//g;
    my $hit =  Bio::LocatableSeq->new('-seq'    => $hs,
                                      '-id'    => $s_nm,
                                      '-start' => $self->hit->start,
                                      '-end'   => $self->hit->end,
                                      '-force_nse' => $s_nm ? 0 : 1,
                                      '-mapping' => [1, $self->{_hit_mapping}]
                                      );
    $aln->add_seq($query);
    $aln->add_seq($hit);
    return $aln;
}

=head2 num_conserved

 Title   : num_conserved
 Usage   : $obj->num_conserved($newval)
 Function: returns the number of conserved residues in the alignment
 Returns : integer
 Args    : integer (optional)


=cut

sub num_conserved{
    my ($self,$value) = @_;

    unless ($self->{_did_presimilar}) {
        $self->_pre_similar_stats;
    }

    if (defined $value) {
        $self->{CONSERVED} = $value;
    }
    return $self->{CONSERVED};
}

=head2 num_identical

 Title   : num_identical
 Usage   : $obj->num_identical($newval)
 Function: returns the number of identical residues in the alignment
 Returns : integer
 Args    : integer (optional)


=cut

sub num_identical{
   my ($self,$value) = @_;

   unless ($self->{_did_presimilar}) {
        $self->_pre_similar_stats;
    }

   if( defined $value) {
       $self->{IDENTICAL} = $value;
   }
   return $self->{IDENTICAL};
}

=head2 rank

 Usage     : $hsp->rank( [string] );
 Purpose   : Get the rank of the HSP within a given Blast hit.
 Example   : $rank = $hsp->rank;
 Returns   : Integer (1..n) corresponding to the order in which the HSP
             appears in the BLAST report.

=cut

sub rank {
    my ($self,$value) = @_;
    if( defined $value) {
        $self->{RANK} = $value;
    }
    return $self->{RANK};
}

=head2 seq_inds

 Title   : seq_inds
 Purpose   : Get a list of residue positions (indices) for all identical
           : or conserved residues in the query or sbjct sequence.
 Example   : @s_ind = $hsp->seq_inds('query', 'identical');
           : @h_ind = $hsp->seq_inds('hit', 'conserved');
           : @h_ind = $hsp->seq_inds('hit', 'conserved-not-identical');
           : @h_ind = $hsp->seq_inds('hit', 'conserved', 1);
 Returns   : List of integers
           : May include ranges if collapse is true.
 Argument  : seq_type  = 'query' or 'hit' or 'sbjct'  (default = query)
           :  ('sbjct' is synonymous with 'hit')
           : class     = 'identical' - identical positions
           :             'conserved' - conserved positions
           :             'nomatch'   - mismatched residue or gap positions
           :             'mismatch'  - mismatched residue positions (no gaps)
           :             'gap'       - gap positions only
           :             'frameshift'- frameshift positions only
           :             'conserved-not-identical' - conserved positions w/o 
           :                            identical residues
           :             The name can be shortened to 'id' or 'cons' unless
           :             the name is ambiguous.  The default value is
           :             'identical'
           :
           : collapse  = boolean, if true, consecutive positions are merged
           :             using a range notation, e.g., "1 2 3 4 5 7 9 10 11"
           :             collapses to "1-5 7 9-11". This is useful for
           :             consolidating long lists. Default = no collapse.
           :
 Throws    : n/a.
 Comments  : For HSPs partially or completely derived from translated sequences
           : (TBLASTN, BLASTX, TBLASTX, or similar), some positional data
           : cannot easily be attributed to a single position (i.e. the 
           : positional data is ambiguous).  In these cases all three codon 
           : positions are reported.  Under these conditions you can check 
           : ambiguous_seq_inds() to determine whether the query, subject, 
           : or both are ambiguous.
           :
See Also   : L<Bio::Search::SearchUtils::collapse_nums()|Bio::Search::SearchUtils>,
             L<Bio::Search::Hit::HitI::seq_inds()|Bio::Search::Hit::HitI>

=cut

sub seq_inds{
   my ($self, $seqType, $class, $collapse) = @_;

   # prepare the internal structures - this is cached so
   # if the strings have not changed we're okay
   $self->_calculate_seq_positions();

   $seqType  ||= 'query';
   $class ||= 'identical';
   $collapse ||= 0;
   $seqType = 'sbjct' if $seqType eq 'hit';
   my $t = lc(substr($seqType,0,1));
   if( $t eq 'q' ) {
       $seqType = 'query';
   } elsif ( $t eq 's' || $t eq 'h' ) {
       $seqType = 'sbjct';
   } else {
       $self->warn("unknown seqtype $seqType using 'query'");
       $seqType = 'query';
   }
   
   $t = lc(substr($class,0,1));

   if( $t eq 'c' ) {
       if( $class =~ /conserved\-not\-identical/ ) {
	   $class = 'conserved';
       } else {
	   $class = 'conservedall';
       }
   } elsif( $t eq 'i' ) {
       $class = 'identical';
   } elsif( $t eq 'n' ) {
       $class = 'nomatch';
   } elsif( $t eq 'm' ) {
       $class = 'mismatch';
   } elsif( $t eq 'g' ) {
       $class = 'gap';
   } elsif( $t eq 'f' ) {
       $class = 'frameshift';    
   } else {
       $self->warn("unknown sequence class $class using 'identical'");
       $class = 'identical';
   }

   ## Sensitive to member name changes.
   $seqType  = "_\L$seqType\E";
   $class = "_\L$class\E";
   my @ary;

   if( $class eq '_gap' ) {
        # this means that we are remapping the gap length that is stored
        # in the hash (for example $self->{'_gapRes_query'} )
        # so we'll return an array which has the values of the position of the
        # of the gap (the key in the hash) + the gap length (value in the
        # hash for this key - 1.
        
        # changing this; since the index is the position prior to the insertion,
        # repeat the position based on the number of gaps inserted
        @ary = map { my @tmp;
                     # position holds number of gaps inserted
                     for my $g (1..$self->{seqinds}{"${class}Res$seqType"}->{$_}) {
                        push @tmp, $_ ;
                     }
                     @tmp}
              sort { $a <=> $b } keys %{ $self->{seqinds}{"${class}Res$seqType"}};
   } elsif( $class eq '_conservedall' ) {
       @ary = sort { $a <=> $b }
       keys %{ $self->{seqinds}{"_conservedRes$seqType"}},
       keys %{ $self->{seqinds}{"_identicalRes$seqType"}},
   }  else {
       @ary = sort { $a <=> $b } keys %{ $self->{seqinds}{"${class}Res$seqType"}};
   }
   require Bio::Search::BlastUtils if $collapse;

   return $collapse ? &Bio::Search::SearchUtils::collapse_nums(@ary) : @ary;
}

=head2 ambiguous_seq_inds

 Title     : ambiguous_seq_inds
 Purpose   : returns a string denoting whether sequence indices for query, 
           : subject, or both are ambiguous
 Returns   : String; 'query', 'subject', 'query/subject', or empty string ''
 Argument  : none
 Comments  : For HSPs partially or completely derived from translated sequences
           : (TBLASTN, BLASTX, TBLASTX, or similar), some positional data
           : cannot easily be attributed to a single position (i.e. the 
           : positional data is ambiguous).  In these cases all three codon 
           : positions are reported when using seq_inds().  Under these
           : conditions you can check ambiguous_seq_inds() to determine whether
           : the query, subject, or both are ambiguous.
See Also   : L<Bio::Search::Hit::HSPI::seq_inds()>

=cut

sub ambiguous_seq_inds {
    my $self = shift;
    $self->_calculate_seq_positions();    
    return $self->{seqinds}{'_warnRes'} if exists $self->{seqinds}{'_warnRes'};
    return $self->{seqinds}{'_warnRes'} = '';
}

=head2 Inherited from L<Bio::SeqFeature::SimilarityPair>

These methods come from L<Bio::SeqFeature::SimilarityPair>

=head2 query

 Title   : query
 Usage   : my $query = $hsp->query
 Function: Returns a SeqFeature representing the query in the HSP
 Returns : L<Bio::SeqFeature::Similarity>
 Args    : [optional] new value to set

=cut

sub query {
    my $self = shift;
    unless ($self->{_created_qff}) {
        $self->_query_seq_feature;
    }
    return $self->SUPER::query(@_);
}

sub feature1 {
    my $self = shift;
    if (! $self->{_finished_new} || $self->{_making_qff}) {
        return $self->{_sim1} if $self->{_sim1};
        $self->{_sim1} = Bio::SeqFeature::Similarity->new();
        return $self->{_sim1};
    }
    unless ($self->{_created_qff}) {
        $self->_query_seq_feature;
    }
    return $self->SUPER::feature1(@_);
}

=head2 hit

 Title   : hit
 Usage   : my $hit = $hsp->hit
 Function: Returns a SeqFeature representing the hit in the HSP
 Returns : L<Bio::SeqFeature::Similarity>
 Args    : [optional] new value to set

=cut

sub hit {
    my $self = shift;
    unless ($self->{_created_sff}) {
        $self->_subject_seq_feature;
    }
    return $self->SUPER::hit(@_);
}

sub feature2 {
    my $self = shift;
    if (! $self->{_finished_new} || $self->{_making_sff}) {
        return $self->{_sim2} if $self->{_sim2};
        $self->{_sim2} = Bio::SeqFeature::Similarity->new();
        return $self->{_sim2};
    }
    unless ($self->{_created_sff}) {
        $self->_subject_seq_feature;
    }
    return $self->SUPER::feature2(@_);
}

=head2 significance

 Title   : significance
 Usage   : $evalue = $obj->significance();
           $obj->significance($evalue);
 Function: Get/Set the significance value
 Returns : numeric
 Args    : [optional] new value to set

=cut 

# Override significance to return the e-value or, if this is
# not defined (WU-BLAST), return the p-value.
sub significance {
    my ($self, $val) = @_;
    if (!defined $self->{SIGNIFICANCE} || defined $val) {
        $self->{SIGNIFICANCE} = defined $val ? $val :
                                defined $self->evalue ?  $self->evalue :
                                defined $self->pvalue ? $$self->pvalue :
                                undef;        
        $self->query->significance($self->{SIGNIFICANCE});
    }
    return $self->{SIGNIFICANCE};
}

=head2 score

 Title   : score
 Usage   : $score = $obj->score();
           $obj->score($value);
 Function: Get/Set the score value
 Returns : numeric
 Args    : [optional] new value to set


=head2 bits

 Title   : bits
 Usage   : $bits = $obj->bits();
           $obj->bits($value);
 Function: Get/Set the bits value
 Returns : numeric
 Args    : [optional] new value to set


=head2 strand

 Title   : strand
 Usage   : $hsp->strand('query')
 Function: Retrieves the strand for the HSP component requested
 Returns : +1 or -1 (0 if unknown)
 Args    : 'hit' or 'subject' or 'sbjct' to retrieve the strand of the subject
           'query' to retrieve the query strand (default)

=cut

=head1 Private methods

=cut

=head2 _calculate_seq_positions

 Title   : _calculate_seq_positions
 Usage   : $self->_calculate_seq_positions
 Function: Internal function
 Returns :
 Args    :


=cut

sub _calculate_seq_positions {
    my ($self,@args) = @_;
    return unless ( $self->{'_sequenceschanged'} );
    $self->{'_sequenceschanged'} = 0;
    my ($seqString, $qseq,$sseq) = ( $self->homology_string(),
                                     $self->query_string(),
                                     $self->hit_string() );
    my ($mlen, $qlen, $slen) = (CORE::length($seqString), CORE::length($qseq), CORE::length($sseq));
    my $qdir = $self->query->strand || 1;
    my $sdir = $self->hit->strand || 1;
    my ($resCount_query, $endpoint_query) = ($qdir <=0) ? ($self->query->end, $self->query->start)
        : ($self->query->start, $self->query->end);
    my ($resCount_sbjct, $endpoint_sbjct) = ($sdir <=0) ? ($self->hit->end, $self->hit->start)
        : ($self->hit->start, $self->hit->end);
    
    my $prog = $self->algorithm;
    
    if( $prog  =~ /FAST|SSEARCH|SMITH-WATERMAN/i ) {
    
        # we infer the end of the regional sequence where the first and last
        # non spaces are in the homology string
        # then we use the HSP->length to tell us how far to read
        # to cut off the end of the sequence
        
        my ($start, $rest) = (0,0);
        if( $seqString =~ /^(\s+)?(.*?)\s*$/ ) {
            ($start, $rest) = ($1 ? CORE::length($1) : 0, CORE::length($2));
        }
    
        $seqString = substr($seqString, $start, $rest);
        $qseq = substr($qseq, $start, $rest);
        $sseq = substr($sseq, $start, $rest);

        # commented out 10/29/08
        # removing frameshift symbols doesn't take into account the following:
        # 1) does not remove the same point in the homology string (get
        # positional errors)
        # 2) adjustments to the overall position in the string due to the
        # frameshift must be taken into consideration (get balancing errors)
        #
        # Frameshifts will be handled directly in the main loop. 
        # --chris
        
        # fasta reports some extra 'regional' sequence information
        # we need to clear out first
        # this seemed a bit insane to me at first, but it appears to
        # work --jason
        
        #$qseq =~ s![\\\/]!!g;
        #$sseq =~ s![\\\/]!!g;
    }

    $self->{seqinds}{'_warnRes'} = '';
    ($self->{_sbjct_offset}, $self->{_query_offset}) = (1,1);
    if($prog =~ /^(?:PSI)?T(BLAST|FAST)(N|X|Y)/oi ) {
        $self->{_sbjct_offset} = 3;
        if ($1 eq 'BLAST' && $2 eq 'X') { #TBLASTX
            $self->{_query_offset} = 3;
            $self->{seqinds}{'_warnRes'} = 'query/subject';
        } else {
            $self->{seqinds}{'_warnRes'} = 'subject';
        }
    } elsif($prog =~ /^(BLAST|FAST)(X|Y|XY)/oi  ) {
        $self->{_query_offset} = 3;
        $self->{seqinds}{'_warnRes'} = 'query';
    }
    my ($qfs, $sfs) = (0,0);
    CHAR_LOOP:
    for my $pos (0..CORE::length($seqString)-1) {
        my @qrange = (0 - $qfs)..($self->{_query_offset} - 1);
        my @srange = (0 - $sfs)..($self->{_sbjct_offset} - 1);
        # $self->debug("QRange:@qrange SRange:@srange\n") if ($qfs || $sfs);
        ($qfs, $sfs) = (0,0);
        my ($mchar, $qchar, $schar) = (
            unpack("x$pos A1",$seqString) || ' ',
            $pos < CORE::length($qseq) ? unpack("x$pos A1",$qseq) : '-',
            $pos < CORE::length($sseq) ? unpack("x$pos A1",$sseq) : '-'
            );
        my $matchtype = '';
        my ($qgap, $sgap) = (0,0);
        if( $mchar eq '+' || $mchar eq '.') {    # conserved
            $self->{seqinds}{_conservedRes_query}{ $resCount_query + ($_ * $qdir) } = 1 for @qrange;
            $self->{seqinds}{_conservedRes_sbjct}{ $resCount_sbjct + ($_ * $sdir) } = 1 for @srange;
            $matchtype = 'conserved';
        } elsif( $mchar eq ':' || $mchar ne ' ' ) { # identical
            $self->{seqinds}{_identicalRes_query}{ $resCount_query + ($_ * $qdir) } = 1 for @qrange;
            $self->{seqinds}{_identicalRes_sbjct}{ $resCount_sbjct + ($_ * $sdir) } = 1 for @srange;
            $matchtype = 'identical';
        } elsif( $mchar eq ' ' ) {  # possible mismatch/nomatch/frameshift
            $qfs = $qchar eq '/'  ?  -1 : # base inserted to match frame
                   $qchar eq '\\' ?   1 : # base deleted to match frame
                   0;
            $sfs = $schar eq '/'  ?  -1 :
                   $schar eq '\\' ?   1 :
                   0;
            if ($qfs) {
                # Frameshifts are tricky.
                
                # '/' indicates that the next residue is shifted back one
                # (-1) frame position and is a deletion of one base (so to
                # correctly align, a base is inserted). That residue should only
                # occupy two sequence positions instead of three.
                
                # '\' indicates that the next residue is shifted forward
                # one (+1) frame position and is an insertion of one base (so to
                # correctly align, a base is removed). That residue should
                # occupy four sequence positions instead of three.
                
                # Note that gaps are not counted across from frameshifts.
                # Frameshift indices are a range of positions starting in the
                # previous sequence position in which the frameshift occurs;
                # they are ambiguous by nature.
                $self->{seqinds}{_frameshiftRes_query}{ $resCount_query - ($_ * $qdir * $qfs) } = $qfs for @qrange;
                $matchtype = "$qfs frameshift-query";
                $sgap = $qgap = 1;
            }
            elsif ($sfs) {
                $self->{seqinds}{_frameshiftRes_sbjct}{ $resCount_sbjct - ($_ * $sdir * $sfs) } = $sfs for @srange;
                $matchtype = "$sfs frameshift-sbcjt";
                $sgap = $qgap = 1;
            }
            elsif ($qchar eq $self->{GAP_SYMBOL}) {
                # gap are counted as being in the immediately preceeding residue
                # position; for translated sequences this is not the start of
                # the previous codon, but the third codon position
                $self->{seqinds}{_gapRes_query}{ $resCount_query - $qdir }++ for @qrange;
                $matchtype = 'gap-query';
                $qgap++;
            }
            elsif ($schar eq $self->{GAP_SYMBOL}) {
                $self->{seqinds}{_gapRes_sbjct}{ $resCount_sbjct - $sdir }++ for @srange;
                $matchtype = 'gap-sbjct';
                $sgap++;
            }
            else {
                # if not a gap or frameshift in either seq, must be mismatch
                $self->{seqinds}{_mismatchRes_query}{ $resCount_query + ($_ * $qdir) } = 1 for @qrange;
                $self->{seqinds}{_mismatchRes_sbjct}{ $resCount_sbjct + ($_ * $sdir) } = 1 for @srange;
                $matchtype = 'mismatch';
            }
            # always add a nomatch unless the current position in the seq is a gap
            if (!$sgap) {
                $self->{seqinds}{_nomatchRes_sbjct}{ $resCount_sbjct + ($_ * $sdir) } = 1 for @srange;
            }
            if (!$qgap) {
                $self->{seqinds}{_nomatchRes_query}{ $resCount_query + ($_ * $qdir) } = 1 for @qrange;
            }
        } else {
            $self->warn("Unknown midline character: [$mchar]");
        }
        # leave in and uncomment for future debugging
        #$self->debug(sprintf("%7d %1s[%1s]%1s %-7d Type: %-20s QOff:%-2d SOff:%-2d\n",
        #                     $resCount_query,
        #                     $qchar,
        #                     $mchar,
        #                     $schar,
        #                     $resCount_sbjct,
        #                     $matchtype,
        #                     ($self->{_query_offset} * $qdir),
        #                     ($self->{_sbjct_offset} * $sdir)));
        $resCount_query += ($qdir * (scalar(@qrange) + $qfs)) if !$qgap;
        $resCount_sbjct += ($sdir * (scalar(@srange) + $sfs)) if !$sgap;
    }
    return 1;
}

=head2 n

See documentation in L<Bio::Search::HSP::HSPI::n()|Bio::Search::HSP::HSPI>

=cut

sub n {
    my $self = shift;
    if(@_) { $self->{'N'} = shift; }
    defined $self->{'N'} ? $self->{'N'} : '';
}

=head2 range

See documentation in L<Bio::Search::HSP::HSPI::range()|Bio::Search::HSP::HSPI>

=cut

sub range {
    my ($self, $seqType) = @_;

    $seqType ||= 'query';
    $seqType = 'sbjct' if $seqType eq 'hit';

    my ($start, $end);
    if( $seqType eq 'query' ) {
        $start = $self->query->start;
        $end = $self->query->end;
    }
    else {
        $start = $self->hit->start;
        $end = $self->hit->end;
    }
    return ($start, $end);
}


=head2 links

 Title   : links
 Usage   : $obj->links($newval)
 Function: Get/Set the Links value (from WU-BLAST)
           Indicates the placement of the alignment in the group of HSPs
 Returns : Value of links
 Args    : On set, new value (a scalar or undef, optional)


=cut

sub links{
    my $self = shift;

    return $self->{LINKS} = shift if @_;
    return $self->{LINKS};
}

=head2 hsp_group

 Title   : hsp_group
 Usage   : $obj->hsp_group($newval)
 Function: Get/Set the Group value (from WU-BLAST)
           Indicates a grouping of HSPs
 Returns : Value of group
 Args    : On set, new value (a scalar or undef, optional)


=cut

sub hsp_group {
    my $self = shift;

    return $self->{HSP_GROUP} = shift if @_;
    return $self->{HSP_GROUP};
}

=head2 hit_features

 Title   : hit_features
 Usage   : $obj->hit_features($newval)
 Function: Get/Set the HSP hit feature string (from some BLAST 2.2.13 text
           output), which is a string of overlapping or nearby features in HSP
           hit
 Returns : Value of hit features, if present
 Args    : On set, new value (a scalar or undef, optional)


=cut

sub hit_features {
    my $self = shift;

    return $self->{HIT_FEATURES} = shift if @_;
    return $self->{HIT_FEATURES};
}

# The cigar string code is written by Juguang Xiao <juguang@fugu-sg.org>

=head1 Brief introduction on cigar string

NOTE: the concept is originally from EnsEMBL docs at
http://may2005.archive.ensembl.org/Docs/wiki/html/EnsemblDocs/CigarFormat.html
Please append to these docs if you have a better definition.

Sequence alignment hits can be stored in a database as ungapped alignments.
This imposes 2 major constraints on alignments:

a) alignments for a single hit record require multiple rows in the database,
and
b) it is not possible to accurately retrieve the exact original alignment.

Alternatively, sequence alignments can be stored as gapped alignments using
the CIGAR line format (where CIGAR stands for Concise Idiosyncratic Gapped
Alignment Report).

In the cigar line format alignments are stored as follows:

M: Match
D: Deletion
I: Insertion

An example of an alignment for a hypthetical protein match is shown below:


Query:   42 PGPAGLP----GSVGLQGPRGLRGPLP-GPLGPPL...

            PG    P    G     GP   R      PLGP

Sbjct: 1672 PGTP*TPLVPLGPWVPLGPSSPR--LPSGPLGPTD...


protein_align_feature table as the following cigar line:

7M4D12M2I2MD7M

=head2 cigar_string

  Name:     cigar_string
  Usage:    $cigar_string = $hsp->cigar_string
  Function: Generate and return cigar string for this HSP alignment
  Args:     No input needed
  Return:   a cigar string

=cut


sub cigar_string {
    my ($self, $arg) = @_;
    $self->warn("this is not a setter") if(defined $arg);

    unless(defined $self->{_cigar_string}){ # generate cigar string
        my $cigar_string = $self->generate_cigar_string($self->query_string, $self->hit_string);
        $self->{_cigar_string} = $cigar_string;
    } # end of unless

    return $self->{_cigar_string};
}

=head2 generate_cigar_string

  Name:     generate_cigar_string
  Usage:    my $cigar_string = Bio::Search::HSP::GenericHSP::generate_cigar_string ($qstr, $hstr);
  Function: generate cigar string from a simple sequence of alignment.
  Args:     the string of query and subject
  Return:   cigar string

=cut

sub generate_cigar_string {
    my ($self, $qstr, $hstr) = @_;
    my @qchars = split //, $qstr;
    my @hchars = split //, $hstr;

    unless(scalar(@qchars) == scalar(@hchars)){
        $self->throw("two sequences are not equal in lengths");
    }

    $self->{_count_for_cigar_string} = 0;
    $self->{_state_for_cigar_string} = 'M';

    my $cigar_string = '';
    for(my $i=0; $i <= $#qchars; $i++){
        my $qchar = $qchars[$i];
        my $hchar = $hchars[$i];
        if($qchar ne $self->{GAP_SYMBOL} && $hchar ne $self->{GAP_SYMBOL}){ # Match
            $cigar_string .= $self->_sub_cigar_string('M');
        }elsif($qchar eq $self->{GAP_SYMBOL}){ # Deletion
            $cigar_string .= $self->_sub_cigar_string('D');
        }elsif($hchar eq $self->{GAP_SYMBOL}){ # Insertion
            $cigar_string .= $self->_sub_cigar_string('I');
        }else{
            $self->throw("Impossible state that 2 gaps on each seq aligned");
        }
    }
    $cigar_string .= $self->_sub_cigar_string('X'); # not forget the tail.
    return $cigar_string;
}

# an internal method to help generate cigar string

sub _sub_cigar_string {
    my ($self, $new_state) = @_;

    my $sub_cigar_string = '';
    if($self->{_state_for_cigar_string} eq $new_state){
        $self->{_count_for_cigar_string} += 1; # Remain the state and increase the counter
    }else{
        $sub_cigar_string .= $self->{_count_for_cigar_string}
            unless $self->{_count_for_cigar_string} == 1;
        $sub_cigar_string .= $self->{_state_for_cigar_string};
        $self->{_count_for_cigar_string} = 1;
        $self->{_state_for_cigar_string} = $new_state;
    }
    return $sub_cigar_string;
}

# needed before seqfeatures can be made
sub _pre_seq_feature {
    my $self = shift;
    my $algo = $self->{ALGORITHM};

    my ($queryfactor, $hitfactor) = (0,0);
    my ($hitmap, $querymap) = (1,1);
    if( $algo =~ /^(?:PSI)?T(?:BLAST|FAST|SW)[NY]/oi ) {
        $hitfactor = 1;
        $hitmap = 3;
    }
    elsif ($algo =~ /^(?:FAST|BLAST)(?:X|Y|XY)/oi || $algo =~ /^P?GENEWISE/oi ) {
        $queryfactor = 1;
        $querymap = 3;
    }
    elsif ($algo =~ /^T(BLAST|FAST|SW)(X|Y|XY)/oi || $algo =~ /^(BLAST|FAST|SW)N/oi || $algo =~ /^WABA|AXT|BLAT|BLASTZ|PSL|MEGABLAST|EXONERATE|SW|SMITH\-WATERMAN|SIM4$/){
        if ($2) {
            $hitmap = $querymap = 3;
        }
        $hitfactor = 1;
        $queryfactor = 1;
    }
    elsif ($algo =~ /^RPS-BLAST/) {
        if ($algo =~ /^RPS-BLAST\(BLASTX\)/) {
            $queryfactor = 1;
            $querymap  = 3;
        }
        $hitfactor = 0;
    }
    else {
        my $stranded = lc substr($self->{STRANDED}, 0,1);
        $queryfactor = ($stranded eq 'q' || $stranded eq 'b') ? 1 : 0;
        $hitfactor = ($stranded eq 'h' || $stranded eq 's' || $stranded eq 'b') ? 1 : 0;
    }
    $self->{_query_factor} = $queryfactor;
    $self->{_hit_factor} = $hitfactor;
    $self->{_hit_mapping} = $hitmap;
    $self->{_query_mapping} = $querymap;
}

# make query seq feature
sub _query_seq_feature {
    my $self = shift;
    $self->{_making_qff} = 1;
    my $qs = $self->{QUERY_START};
    my $qe = $self->{QUERY_END};
    unless (defined $self->{_query_factor}) {
        $self->_pre_seq_feature;
    }
    my $queryfactor = $self->{_query_factor};

    unless( defined $qe && defined $qs ) { $self->throw("Did not specify a Query End or Query Begin"); }

    my $strand;
    if ($qe > $qs) {  # normal query: start < end
        if ($queryfactor) {
            $strand = 1;
        }
        else {
            $strand = undef;
        }
    }
    else {
        if ($queryfactor) {
            $strand = -1;
        }
        else {
            $strand = undef;
        }
        ($qs,$qe) = ($qe,$qs);
    }

    # Note: many of these data are not query- and hit-specific.
    # Only start, end, name, length are.
    # We could be more efficient by only storing this info once.
    # steve chervitz --- Sat Apr  5 00:55:07 2003

    my $sim1 = $self->{_sim1} || Bio::SeqFeature::Similarity->new();
    $sim1->start($qs);
    $sim1->end($qe);
    $sim1->significance($self->{EVALUE});
    $sim1->bits($self->{BITS});
    $sim1->score($self->{SCORE});
    $sim1->strand($strand);
    $sim1->seq_id($self->{QUERY_NAME});
    $sim1->seqlength($self->{QUERY_LENGTH});
    $sim1->source_tag($self->{ALGORITHM});
    $sim1->seqdesc($self->{QUERY_DESC});
    $sim1->add_tag_value('meta', $self->{META}) if $self->can('meta');
    # to determine frame from something like FASTXY which doesn't
    # report the frame
    my $qframe = $self->{QUERY_FRAME};
    
    if (defined $strand && !defined $qframe && $queryfactor) {
        $qframe = ( $qs % 3 ) * $strand;
    }
    elsif (!defined $strand) {
        $qframe = 0;
    }
    
    if( $qframe =~ /^([+-])?([0-3])/ ) {
        my $dir = $1 || '+';
        if($qframe && (($dir eq '-' && $strand >= 0) || ($dir eq '+' && $strand <= 0)) ) {
            $self->warn("Query frame ($qframe) did not match strand of query ($strand)");
        }
        $qframe = $2 != 0 ? $2 - 1 : $2;
    }  else {
        $self->warn("Unknown query frame ($qframe)");
        $qframe = 0;
    }
    
    $sim1->frame($qframe);
    $self->SUPER::feature1($sim1);

    $self->{_created_qff} = 1;
    $self->{_making_qff} = 0;
}

# make subject seq feature
sub _subject_seq_feature {
    my $self = shift;
    $self->{_making_sff} = 1;
    my $hs = $self->{HIT_START};
    my $he = $self->{HIT_END};
    unless (defined $self->{_hit_factor}) {
        $self->_pre_seq_feature;
    }
    my $hitfactor = $self->{_hit_factor};

    unless( defined $he && defined $hs ) { $self->throw("Did not specify a Hit End or Hit Begin"); }

    my $strand;
    if ($he > $hs) { # normal subject
        if ($hitfactor) {
            $strand = 1;
        }
        else {
            $strand = undef;
        }
    }
    else {
        if ($hitfactor) {
            $strand = -1;
        }
        else {
            $strand = undef;
        }
        ($hs,$he) = ( $he,$hs); # reverse subject: start bigger than end
    }

    my $sim2 = $self->{_sim2} || Bio::SeqFeature::Similarity->new();
    $sim2->start($hs);
    $sim2->end($he);
    $sim2->significance($self->{EVALUE});
    $sim2->bits($self->{BITS});
    $sim2->score($self->{SCORE});
    $sim2->strand($strand);
    $sim2->seq_id($self->{HIT_NAME});
    $sim2->seqlength($self->{HIT_LENGTH});
    $sim2->source_tag($self->{ALGORITHM});
    $sim2->seqdesc($self->{HIT_DESC});
    $sim2->add_tag_value('meta', $self->{META}) if $self->can('meta');
    my $hframe = $self->{HIT_FRAME};
    
    if (defined $strand && !defined $hframe && $hitfactor) {
        $hframe = ( $hs % 3 ) * $strand;
    }
    elsif (!defined $strand) {
        $hframe = 0;
    }
    
    if( $hframe =~ /^([+-])?([0-3])/ ) {
        my $dir = $1 || '+';
        if($hframe && (($dir eq '-' && $strand >= 0) || ($dir eq '+' && $strand <= 0)) ) {
            $self->warn("Subject frame ($hframe) did not match strand of subject ($strand)");
        }
        $hframe = $2 != 0 ? $2 - 1 : $2;
    }  else {
        $self->warn("Unknown subject frame ($hframe)");
        $hframe = 0;
    }
    
    $sim2->frame($hframe);
    $self->SUPER::feature2($sim2);

    $self->{_created_sff} = 1;
    $self->{_making_sff} = 0;
}

# before calling the num_* methods
sub _pre_similar_stats {
    my $self = shift;
    my $identical = $self->{IDENTICAL};
    my $conserved = $self->{CONSERVED};
    my $percent_id = $self->{PERCENT_IDENTITY};

    if (! defined $identical) {
        if (! defined $percent_id) {
            $self->warn("Did not defined the number of identical matches or overall percent identity in the HSP; assuming 0");
            $identical = 0;
        }
        else {
            $identical = sprintf("%.0f",$percent_id * $self->{HSP_LENGTH});
        }
    }

    if (! defined $conserved) {
        $self->warn("Did not define the number of conserved matches in the HSP; assuming conserved == identical ($identical)")
            if( $self->{ALGORITHM} !~ /^((FAST|BLAST)N)|EXONERATE|SIM4|AXT|PSL|BLAT|BLASTZ|WABA/oi);
        $conserved = $identical;
    }
    $self->{IDENTICAL} = $identical;
    $self->{CONSERVED} = $conserved;
    $self->{_did_presimilar} = 1;
}

# before calling the frac_* methods

sub _pre_frac {
    my $self = shift;
    my $hsp_len = $self->{HSP_LENGTH};
    my $hit_len = $self->{HIT_LENGTH};
    my $query_len = $self->{QUERY_LENGTH};

    my $identical = $self->num_identical;
    my $conserved = $self->num_conserved;

    $self->{_did_prefrac} = 1;
    my $logical;
    if( $hsp_len ) {
        $self->length('total', $hsp_len);
        $logical = $self->_logical_length('total');
        $self->frac_identical( 'total', $identical / $hsp_len);
        $self->frac_conserved( 'total', $conserved / $hsp_len);
    }
    if( $hit_len ) {
        $logical = $self->_logical_length('hit');
        $self->frac_identical( 'hit', $identical / $logical);
        $self->frac_conserved( 'hit', $conserved / $logical);
    }
    if( $query_len ) {
        $logical = $self->_logical_length('query');
        $self->frac_identical( 'query', $identical / $logical) ;
        $self->frac_conserved( 'query', $conserved / $logical);
    }
}

# before calling gaps()
# This relies first on passed parameters (parser-dependent), then on gaps
# calculated by seq_inds() (if set), then falls back to directly checking
# for '-' as a last resort  

sub _pre_gaps {
    my $self = shift;
    my $query_gaps = $self->{QUERY_GAPS};
    my $query_seq = $self->{QUERY_SEQ};
    my $hit_gaps = $self->{HIT_GAPS};
    my $hit_seq = $self->{HIT_SEQ};
    my $gaps = $self->{HSP_GAPS};

    $self->{_did_pregaps} = 1; # well, we're in the process; avoid recursion
    if( defined $query_gaps ) {
        $self->gaps('query', $query_gaps);
    } elsif( defined $query_seq ) {
        my $qg = (defined $self->{'_query_offset'}) ? $self->seq_inds('query','gaps') : scalar( $query_seq =~ tr/\-//);
        my $offset = $self->{'_query_offset'} || 1;
        $self->gaps('query', $qg/$offset);
    }
    if( defined $hit_gaps ) {
        $self->gaps('hit', $hit_gaps);
    } elsif( defined $hit_seq ) {
        my $hg = (defined $self->{'_sbjct_offset'}) ? $self->seq_inds('hit','gaps') : scalar( $hit_seq =~ tr/\-//);
        my $offset = $self->{'_sbjct_offset'} || 1;
        $self->gaps('hit', $hg/$offset);
    }
    if( ! defined $gaps ) {
        $gaps = $self->gaps("query") + $self->gaps("hit");
    }
    $self->gaps('total', $gaps);
}

# before percent_identity
sub _pre_pi {
    my $self = shift;
    $self->{_did_prepi} = 1;
    $self->percent_identity($self->{PERCENT_IDENTITY} || $self->frac_identical('total')*100) if( $self->{HSP_LENGTH} > 0 );
}

1;