Bio-FastParsers

 view release on metacpan or  search on metacpan

lib/Bio/FastParsers/Blast/Table.pm  view on Meta::CPAN

    my $self = shift;

    open my $fh, '<', $self->file;      # autodie
    return sub { <$fh> };               # return closure
}

## use critic

my @attrs = qw(
    query_id hit_id
    percent_identity hsp_length mismatches gaps
    query_from  query_to
      hit_from    hit_to
    evalue bit_score
    query_strand
      hit_strand
    query_start query_end
      hit_start   hit_end
);      # DID try to use MOP to get HSP attrs but order was not preserved

lib/Bio/FastParsers/Blast/Table/Hsp.pm  view on Meta::CPAN

    is       => 'ro',
    isa      => 'Str',
    required => 1,
) for qw(query_id hit_id);

has $_ => (
    is       => 'ro',
    isa      => 'Num',
    required => 1,
) for qw(
    percent_identity hsp_length mismatches gaps
    query_from  query_to
      hit_from    hit_to
    query_strand
      hit_strand
    query_start query_end
      hit_start   hit_end
);

has $_ => (
    is       => 'ro',

lib/Bio/FastParsers/Blast/Table/Hsp.pm  view on Meta::CPAN

Returns the id of the query sequence.

This method does not accept any arguments.

=head2 hit_id

Returns the id of the hit (or subject) sequence.

This method does not accept any arguments.

=head2 percent_identity

Returns the identity (in percents) of the HSP.

This method does not accept any arguments.

=head2 hsp_length

Returns the length (in nt or aa) of the HSP.

This method does not accept any arguments.

=head2 mismatches

lib/Bio/FastParsers/Blast/Xml/Hsp.pm  view on Meta::CPAN

    return shift->query_coverage
}


sub scov {
    return shift->subject_coverage
}


sub pident {
    return shift->percentage_identity
}


sub ppos {
    return shift->percentage_positive
}



sub query_len {
    return shift->_parent->query_len
}

lib/Bio/FastParsers/Blast/Xml/Hsp.pm  view on Meta::CPAN

sub subject_coverage {
    my $self = shift;
    return sprintf("%.1f",
        100 * ( $self->hit_end - $self->hit_start + 1 ) / $self->hit_len
    );
}


sub percentage_identity {
    my $self = shift;
    return sprintf("%.1f", 100 * ( $self->identity / $self->align_len ) );
}


sub percentage_positive {
    my $self = shift;
    return sprintf("%.1f", 100 * ( $self->positive / $self->align_len ) );
}


__PACKAGE__->meta->make_immutable;
1;

__END__

lib/Bio/FastParsers/Blast/Xml/Hsp.pm  view on Meta::CPAN

=head2 subject_coverage

Returns the subject (hit) coverage of the HSP.

    # $hsp is a Bio::FastParsers::Blast::Xml::Hsp
    my $subject_coverage = $hsp->subject_coverage;

This method does not accept any arguments.

=head2 percentage_identity

Returns the percentage of identity of the HSP.

    # $hsp is a Bio::FastParsers::Blast::Xml::Hsp
    my $percentage_identity = $hsp->percentage_identity;

This method does not accept any arguments.

=head2 percentage_positive

Returns the percentage of positive matches of the HSP.

    # $hsp is a Bio::FastParsers::Blast::Xml::Hsp
    my $percentage_positive = $hsp->percentage_positive;

This method does not accept any arguments.

=head1 ALIASES

=head2 expect

Alias for C<evalue> method. For API consistency.

=head2 qcov

Alias for C<query_coverage> method. For API consistency.

=head2 scov

Alias for C<subject_coverage> method. For API consistency.

=head2 pident

Alias for C<percentage_identity> method. For API consistency.

=head2 ppos

Alias for C<percentage_positive> method. For API consistency.

=head2 query_len

Alias for C<query_len> method in Hit object. For API completeness.

=head2 hit_len

Alias for C<len> method in Hit object. For API completeness.

=head1 AUTHOR

lib/Bio/FastParsers/Uclust.pm  view on Meta::CPAN

    LINE:
    while (my $line = <$in>) {
        chomp $line;
        my ($type, @fields) = split /\t/xms, $line;

        # https://www.drive5.com/usearch/manual/opt_uc.html
        # Field Description
        # - Record type S, H, C or N (see table below).
        # 0 Cluster number (0-based).
        # 1 Sequence length (S, N and H) or cluster size (C).
        # 2 For H records, percent identity with target.
        # 3 For H records, the strand: + or - for nucleotides, . for proteins.
        # 4 Not used, parsers should ignore this field. Included for backwards compatibility.
        # 5 Not used, parsers should ignore this field. Included for backwards compatibility.
        # 6 Compressed alignment or the symbol '=' (equals sign). The = indicates that the query is 100% identical to the target sequence (field 10).
        # 7 Label of query sequence (always present).
        # 8 Label of target sequence (H records only).

        if    ($type eq 'C') {
            push @{ $members_for{ $fields[7] } }, ();
        }

t/blast_table.t  view on Meta::CPAN

    ]
);


sub check_hsps {
    my $infile = shift;
    my $exp_hsps_ref = shift;

    my @methods = qw(
        query_id hit_id
        percent_identity hsp_length mismatches gaps
        query_from  query_to
          hit_from    hit_to
        evalue bit_score
        query_strand
          hit_strand
        query_start query_end
          hit_start   hit_end
    );

    ok my $report = $class->new( file => $infile ), 'Blast::Table constructor';

t/blast_table.t  view on Meta::CPAN

sub check_hits {
    my $infile = shift;
    my $exp_hsps_ref = shift;

    my $report = $class->new( file => $infile );

    while ( my $exp_hsp = shift @{ $exp_hsps_ref } ) {
        my $hsp = $report->next_hit;
        cmp_deeply [ $hsp->query_id, $hsp->hit_id, $hsp->percent_identity + 0 ],
            $exp_hsp, 'got expected next hit';
    }

    return;
}


sub check_queries {
    my $infile = shift;
    my $exp_hsps_ref = shift;

    my $report = $class->new( file => $infile );

    while ( my $exp_hsp = shift @{ $exp_hsps_ref } ) {
        my $hsp = $report->next_query;
        cmp_deeply [ $hsp->query_id, $hsp->hit_id, $hsp->percent_identity + 0 ],
            $exp_hsp, 'got expected next query';
    }

    return;
}

# TODO: test multi-query reports

done_testing;

t/blast_xml.t  view on Meta::CPAN

    ];
    
    my @hsp_data = map {
        $_->query_strand, $_->hit_strand,
        $_->query_start, $_->query_end,
        $_->hit_start, $_->hit_end,
        $_->query_frame, $_->query_from, $_->query_to,
        $_->hit_frame, $_->hit_from, $_->hit_to,
        $_->query_len, $_->hit_len,
        $_->qcov, $_->subject_coverage,
        $_->pident, $_->percentage_positive
    } map { $_->get_hsp(27) } $bo->get_iteration(0)->get_hit(0);

    cmp_deeply \@hsp_data, $hsp_data_ref,
        'got expected coordinates for hsp 27';
}


sub check_file {
	my %args = @_;