use Bio::LocatableSeq;
    my $seq = Bio::LocatableSeq->new(-seq => "CAGT-GGT",
                    -id  => "seq1",
                    -start => 1,
                    -end   => 7);

    # a normal sequence object
    $locseq->seq();
    $locseq->id();

    # has start,end points
    $locseq->start();
    $locseq->end();

    # inherits off RangeI, so range operations possible

    my ($class, @args) = @_;
    my $self = $class->SUPER::new(@args);

    my ($start,$end,$strand, $mapping, $fs, $nse) =
    $self->_rearrange( [qw(START
                        END
                        STRAND
                        MAPPING
                        FRAMESHIFTS
                        FORCE_NSE
                        )],
               @args);
    
    $mapping ||= [1,1];
    $self->mapping($mapping);
    $nse || 0;
    $self->force_nse($nse);
    defined $fs    && $self->frameshifts($fs);
    defined $start && $self->start($start);
    defined $end   && $self->end($end);
    defined $strand && $self->strand($strand);

    return $self; # success - we hope!

    my $self = shift;
    if( @_ ) {
        my $value = shift;
        $self->{'start'} = $value;
    }
    return $self->{'start'} if defined $self->{'start'};
    return 1                if $self->seq;
    return;

    my $self = shift;
    if( @_ ) {
        my $value = shift;
        my $st = $self->start;
        # start of 0 usually means the sequence is all gaps but maps to
        # other sequences in an alignment
        if ($self->seq && $st != 0 ) {
            my $len = $self->_ungapped_len;
            my $calend = $st + $len - 1;
            my $id = $self->id || 'unknown';
            if ($calend != $value) {
                $self->warn("In sequence $id residue count gives end value ".
                "$calend.\n  Overriding value [$value] with value $calend for ".
                "Bio::LocatableSeq::end().\n".$self->seq);
                $value = $calend;
            }
        }
        $self->{'end'} = $value;
    }

    if (defined $self->{'end'}) {
        return $self->{'end'}
    } elsif ( my $len = $self->_ungapped_len) {
        return $len + $self->start - 1;
    } else {
        return;
    }
}

# changed 08.10.26 to return ungapped length, not the calculated end
# of the sequence

    my $self = shift;
    return unless my $string = $self->seq;
    my ($map_res, $map_coord) = $self->mapping;
    my $offset = 0;
    if (my %data = $self->frameshifts) {
        map {$offset += $_} values %data;
    }
    $string =~ s{[$GAP_SYMBOLS$FRAMESHIFT_SYMBOLS]+}{}g;
    return CORE::length($string)/($map_res/$map_coord) + $offset/($map_coord/$map_res);
}

#sub length {
#    my $self = shift;
#    return unless my $string = $self->seq;
#    $string =~ s{[$GAP_SYMBOLS$FRAMESHIFT_SYMBOLS]+}{}g;
#    return CORE::length($string);
#

   my $self = shift;
   if( @_ ) {
        my $value = shift;
        $self->{'strand'} = $value;
    }
    return $self->{'strand'};

    my $self = shift;
    if( @_ ) {
        my @mapping = (ref($_[0]) eq 'ARRAY') ? @{$_[0]} : @_;
        $self->throw("Must pass two values (# residues mapped to # positions)")
            if @mapping != 2;
        if ((grep {$_ != 1 && $_ != 3} @mapping) || ($mapping[0] == 3 && $mapping[1] == 3)) {
            $self->throw("Mapping values other than 1 or 3 are not currently supported")
        }
        $self->{'_mapping'} =\@ mapping;
    }
    $self->throw('Mapping for LocatableSeq not set') if !exists $self->{'_mapping'};
    return @{ $self->{'_mapping'} };

    my $self = shift;
    if( @_ ) {
        if (ref $_[0] eq 'HASH') {
            $self->{_frameshifts} = $_[0];
        } else {
            # assume this is a full list to be converted to a hash
            $self->{_frameshifts} =\% {@_} # coerce into hash ref
        }
    }
    (defined $self->{_frameshifts} && ref $self->{_frameshifts} eq 'HASH') ?
        return %{$self->{_frameshifts}} : return ();

   my ($self,$char1,$char2) = @_;

   $char1 ||= "/";
   $char2 ||= "-";
   
   my ($id, $st, $end, $strand)  = ($self->id(), $self->start(),
                                    $self->end(), $self->strand || 0);
   
   if ($self->force_nse) {
        $id  ||= '';
        $st  ||= 0;
        $end ||= 0;
   }
   
   $self->throw("Attribute id not set") unless defined($id);
   $self->throw("Attribute start not set") unless defined($st);
   $self->throw("Attribute end not set") unless defined($end);
   
   if ($strand && $strand == -1) {
      ($st, $end) = ($end, $st);
   }
   
   #Stockholm Rfam includes version if present so it is optional
   my $v = $self->version ? '.'.$self->version : '';
   return join('',$id, $v, $char1, $st, $char2, $end);

    my ($self, $flag) = @_;
    if (defined $flag) {
        $flag ? (return $self->{'_force_nse'} = 1) : (return $self->{'_force_nse'} = 0);
    }
    return $self->{'_force_nse'};

    my ($self,$char) = @_;
    my ($seq, $count) = (undef, 0);

    # default gap characters
    $char ||= $GAP_SYMBOLS;

    $self->warn("I hope you know what you are doing setting gap to [$char]")
        unless $char =~ /[$GAP_SYMBOLS]/;

    $seq = $self->seq;
    return 0 unless $seq; # empty sequence does not have gaps

    $seq =~ s/^([$char]+)//;
    $seq =~ s/([$char]+)$//;
    while ( $seq =~ /[$char]+/g ) {
        $count++;
    }

    return $count;

    my ($self, $resnumber) = @_;

    $self->throw("Residue number has to be a positive integer, not [$resnumber]")
    unless $resnumber =~ /^\d+$/ and $resnumber > 0;

    if ($resnumber >= $self->start() and $resnumber <= $self->end()) {
	my @chunks;
	my $column_incr;
	my $current_column;
	my $current_residue = $self->start - 1;
	my $seq = $self->seq;
	my $strand = $self->strand || 0;

	if ($strand == -1) {
#	    @chunks = reverse $seq =~ m/[^\.\-]+|[\.\-]+/go;
	    @chunks = reverse $seq =~ m/[$RESIDUE_SYMBOLS]+|[$GAP_SYMBOLS]+/go;
	    $column_incr = -1;
	    $current_column = (CORE::length $seq) + 1;
	}
	else {
#	    @chunks = $seq =~ m/[^\.\-]+|[\.\-]+/go;
	    @chunks = $seq =~ m/[$RESIDUE_SYMBOLS]+|[$GAP_SYMBOLS]+/go;
	    $column_incr = 1;
	    $current_column = 0;
	}

	while (my $chunk = shift @chunks) {
#	    if ($chunk =~ m|^[\.\-]|o) {
	    if ($chunk =~ m|^[$GAP_SYMBOLS]|o) {
		$current_column += $column_incr * CORE::length($chunk);
	    }
	    else {
		if ($current_residue + CORE::length($chunk) < $resnumber) {
		    $current_column += $column_incr * CORE::length($chunk);
		    $current_residue += CORE::length($chunk);
		}
		else {
		    if ($strand == -1) {
			$current_column -= $resnumber - $current_residue;
		    }
		    else {
			$current_column += $resnumber - $current_residue;
		    }
		    return $current_column;
		}
	    }
	}
    }

    $self->throw("Could not find residue number $resnumber");

    my ($self, $column) = @_;

    $self->throw("Column number has to be a positive integer, not [$column]")
    unless $column =~ /^\d+$/ and $column > 0;
    $self->throw("Column number [$column] is larger than".
         " sequence length [". $self->length. "]")
    unless $column <= $self->length;

    my ($loc);
    my $s = $self->subseq(1,$column);
    $s =~ s/[^a-zA-Z\*]//g;

    my $pos = CORE::length $s;

    my $start = $self->start || 0 ;
    my $strand = $self->strand() || 1;
    my $relative_pos = ($strand == -1)
        ? ($self->end - $pos + 1)
    : ($pos + $start - 1);
    if ($self->subseq($column, $column) =~ /[a-zA-Z\*]/ ) {
    $loc = Bio::Location::Simple->new
        (-start  => $relative_pos,
         -end    => $relative_pos,
         -strand => 1,
         );
    } elsif ($pos == 0 and $self->start == 1) {
    } else {
      my ($start,$end) = ($relative_pos, $relative_pos + $strand);
      if ($strand == -1) {
    ($start,$end) = ($end,$start);
      }
    $loc = Bio::Location::Simple->new
        (-start         => $start,
         -end           => $end,
         -strand        => 1,
         -location_type => 'IN-BETWEEN'
         );
    }
    return $loc;

    my ($self) = @_;
    # since we don't know whether sequences without 1 => 1 correlation can be
    # revcom'd, kick back
    if (grep {$_ != 1} $self->mapping) {
        $self->warn('revcom() not supported for sequences with mapped values of > 1');
        return;
    }
    my $new = $self->SUPER::revcom;
    $new->strand($self->strand * -1) if $self->strand;
    $new->start($self->start) if $self->start;
    $new->end($self->end) if $self->end;
    return $new;

    my ($self, $start, $end) = @_;
    my $new = $self->SUPER::trunc($start, $end);
    $new->strand($self->strand);

    # end will be automatically calculated
    $start = $end if $self->strand && $self->strand == -1;

    $start = $self->location_from_column($start);
    $start ? ($start = $start->end) : ($start = 1);
    $new->start($start) if $start;

    return $new;

    my ($self,$seqstr) = @_;
    if( ! defined $seqstr ){ $seqstr = $self->seq(); }
    return 0 unless( defined $seqstr);
    
    if((CORE::length($seqstr) > 0) &&
       ($seqstr !~ /^([$MATCHPATTERN]+)$/)) {
        $self->warn("seq doesn't validate with [$MATCHPATTERN], mismatch is " .
            join(",",($seqstr =~ /([^$MATCHPATTERN]+)/g)));
        return 0;
    }
    return 1;
}

################## DEPRECATED METHODS ##################

	my $self = shift;
	$self->deprecated(-warn_version => 1.0069,
					  -throw_version => 1.0075,
                      -message => 'Use of method no_gaps() is deprecated, use num_gaps() instead');
    $self->num_gaps(@_);

	my $self = shift;
	$self->deprecated(-warn_version => 1.0069,
					  -throw_version => 1.0075,
                      -message => 'Use of method no_sequences() is deprecated, use num_sequences() instead');
    $self->num_sequences(@_);
}

1;

  bioperl-l@bioperl.org                  - General discussion
  http://bioperl.org/wiki/Mailing_lists  - About the mailing lists

  https://redmine.open-bio.org/projects/bioperl/

 Title     : no_gaps
 Usage     : $self->no_gaps('.')
 Function  : Gets number of gaps in the sequence. The count excludes
             leading or trailing gap characters.

             Valid bioperl sequence characters are [A-Za-z\-\.\*]. Of
             these, '.' and '-' are counted as gap characters unless an
             optional argument specifies one of them.

 Returns   : number of internal gaps in the sequence.
 Args      : a gap character (optional)
 Status    : Deprecated (in favor of num_gaps())