[Bioperl-l] about the FgeneSH parser
Magic Fang
fangl at genomics.org.cn
Tue Nov 11 20:32:32 EST 2003
hi, in the attachment is the fgenesh packages drived from genscan
module, i fixed some problem for GFF output, include add the seq_id etc.
there are some codes i marked. please check it.
Magic Fang
Todd Harris wrote:
>>Hi,
>>
>>Sure, but...
>>
>>How does it compare with Bio::EnsEMBL::Pipeline::Runnable module?
>>
>>http://www.ensembl.org/Docs/Pdoc/ensembl-pipeline/modules/Bio/EnsEMBL/Pipeline
>>/Runnable/Fgenesh.html
>>
>>or this:
>>
>>http://www2.toddot.net:8081/research/scripts/parsers/fgenesh/
>>
>>
>
>Ha! That's my total FGENESH parser hack! The power of google never ceases
>to amaze.
>
>todd
>
>
>
>>-Heikki
>>
>>On Mon, 2003-11-10 at 08:11, Magic Fang wrote:
>>
>>
>>>i have developed a fgenesh parser these days, it is base on genscan
>>>package, current version is 0.1. i think fgenesh is the powerful gene
>>>finding program. if i can add it to the bioperl code tree?
>>>
>>>_______________________________________________
>>>Bioperl-l mailing list
>>>Bioperl-l at portal.open-bio.org
>>>http://portal.open-bio.org/mailman/listinfo/bioperl-l
>>>
>>>
>
>
>
>
>
-------------- next part --------------
# $Id: FgeneSH.pm,v 1.21 2002/10/08 08:38:32 lapp Exp $
#
# BioPerl module for Bio::Tools::FgeneSH
#
# Cared for by Magic Fang <fangl at genomics.org.cn>
#
# Copyright Magic Fang
#
# You may distribute this module under the same terms as perl itself
# POD documentation - main docs before the code
=head1 NAME
Bio::Tools::FgeneSH - Results of one FgeneSH run
=head1 SYNOPSIS
$fgenesh = Bio::Tools::FgeneSH->new(-file => 'result.fgenesh');
# filehandle:
$fgenesh = Bio::Tools::FgeneSH->new( -fh => \*INPUT );
# parse the results
# note: this class is-a Bio::Tools::AnalysisResult which implements
# Bio::SeqAnalysisParserI, i.e., $fgenesh->next_feature() is the same
while($gene = $fgenesh->next_prediction()) {
# $gene is an instance of Bio::Tools::Prediction::Gene, which inherits
# off Bio::SeqFeature::Gene::Transcript.
#
# $gene->exons() returns an array of
# Bio::Tools::Prediction::Exon objects
# all exons:
@exon_arr = $gene->exons();
# initial exons only
@init_exons = $gene->exons('Initial');
# internal exons only
@intrl_exons = $gene->exons('Internal');
# terminal exons only
@term_exons = $gene->exons('Terminal');
# singleton exons:
($single_exon) = $gene->exons();
}
# essential if you gave a filename at initialization (otherwise the file
# will stay open)
$fgenesh->close();
=head1 DESCRIPTION
The FgeneSH module provides a parser for FgeneSH gene structure prediction
output. It parses one gene prediction into a Bio::SeqFeature::Gene::Transcript-
derived object.
This module also implements the Bio::SeqAnalysisParserI interface, and thus
can be used wherever such an object fits. See L<Bio::SeqAnalysisParserI>.
=head1 FEEDBACK
=head2 Mailing Lists
User feedback is an integral part of the evolution of this and other
Bioperl modules. Send your comments and suggestions preferably to one
of the Bioperl mailing lists. Your participation is much appreciated.
bioperl-l at bioperl.org - General discussion
http://bio.perl.org/MailList.html - About the mailing lists
=head2 Reporting Bugs
Report bugs to the Bioperl bug tracking system to help us keep track
the bugs and their resolution. Bug reports can be submitted via email
or the web:
bioperl-bugs at bio.perl.org
http://bio.perl.org/bioperl-bugs/
=head1 AUTHOR - Hilmar Lapp
Email hlapp at gmx.net
Describe contact details here
=head1 APPENDIX
The rest of the documentation details each of the object methods. Internal methods are usually preceded with a _
=cut
# Let the code begin...
package Bio::Tools::FgeneSH;
use vars qw(@ISA);
use strict;
use Symbol;
use Bio::Root::Root;
use Bio::Tools::AnalysisResult;
use Bio::Tools::Prediction::Gene;
use Bio::Tools::Prediction::Exon;
@ISA = qw(Bio::Tools::AnalysisResult);
sub _initialize_state {
my ($self, at args) = @_;
# first call the inherited method!
$self->SUPER::_initialize_state(@args);
# our private state variables
$self->{'_preds_parsed'} = 0;
$self->{'_has_cds'} = 0;
# array of pre-parsed predictions
$self->{'_preds'} = [];
# seq stack
$self->{'_seqstack'} = [];
}
=head2 analysis_method
Usage : $fgenesh->analysis_method();
Purpose : Inherited method. Overridden to ensure that the name matches
/fgenesh/i.
Returns : String
Argument : n/a
=cut
#-------------
sub analysis_method {
#-------------
my ($self, $method) = @_;
if($method && ($method !~ /fgenesh/i)) {
$self->throw("method $method not supported in " . ref($self));
}
return $self->SUPER::analysis_method($method);
}
=head2 next_feature
Title : next_feature
Usage : while($gene = $fgenesh->next_feature()) {
# do something
}
Function: Returns the next gene structure prediction of the FgeneSH result
file. Call this method repeatedly until FALSE is returned.
The returned object is actually a SeqFeatureI implementing object.
This method is required for classes implementing the
SeqAnalysisParserI interface, and is merely an alias for
next_prediction() at present.
Example :
Returns : A Bio::Tools::Prediction::Gene object.
Args :
=cut
sub next_feature {
my ($self, at args) = @_;
# even though next_prediction doesn't expect any args (and this method
# does neither), we pass on args in order to be prepared if this changes
# ever
return $self->next_prediction(@args);
}
=head2 next_prediction
Title : next_prediction
Usage : while($gene = $fgenesh->next_prediction()) {
# do something
}
Function: Returns the next gene structure prediction of the FgeneSH result
file. Call this method repeatedly until FALSE is returned.
Example :
Returns : A Bio::Tools::Prediction::Gene object.
Args :
=cut
sub next_prediction {
my ($self) = @_;
my $gene;
# if the prediction section hasn't been parsed yet, we do this now
$self->_parse_predictions() unless $self->_predictions_parsed();
# get next gene structure
$gene = $self->_prediction();
# here i patch the GenScan package bug when we meet a result just contain
# promoter or polyA records
# Magic Fang(fangl at genomics.org.cn) BGI, 2003-11-10
if($gene && $gene->exons>0) {
# fill in predicted protein, and if available the predicted CDS
#
my ($id, $seq);
# use the seq stack if there's a seq on it
my $seqobj = pop(@{$self->{'_seqstack'}});
if(! $seqobj) {
# otherwise read from input stream
($id, $seq) = $self->_read_fasta_seq();
# there may be no sequence at all, or none any more
if($id && $seq) {
$seqobj = Bio::PrimarySeq->new('-seq' => $seq,
'-display_id' => $id,
'-alphabet' => "protein");
}
}
if($seqobj) {
# check that prediction number matches the prediction number
# indicated in the sequence id (there may be incomplete gene
# predictions that contain only signals with no associated protein
# and CDS, like promoters, poly-A sites etc)
$gene->primary_tag() =~ /[^0-9]([0-9]+)$/;
my $prednr = $1;
if($seqobj->display_id() !~ /FGENESH\:\s+$prednr\s+/) {
# this is not our sequence, so push back for next prediction
push(@{$self->{'_seqstack'}}, $seqobj);
} else {
$gene->predicted_protein($seqobj);
# CDS prediction, too?
if($self->_has_cds()) {
($id, $seq) = $self->_read_fasta_seq();
$seqobj = Bio::PrimarySeq->new('-seq' => $seq,
'-display_id' => $id,
'-alphabet' => "dna");
$gene->predicted_cds($seqobj);
}
}
}
}
return $gene;
}
=head2 _parse_predictions
Title : _parse_predictions()
Usage : $obj->_parse_predictions()
Function: Parses the prediction section. Automatically called by
next_prediction() if not yet done.
Example :
Returns :
=cut
sub _parse_predictions {
my ($self) = @_;
my %exontags = ('CDSf' => 'Initial',
'CDSi' => 'Internal',
'CDSl' => 'Terminal',
'CDSo' => '');
my $gene;
my $seqname;
while(defined($_ = $self->_readline())) {
if(/^\s*(\d+)\s*[\+|\-]/) { # catch the gene region line
# exon or signal
my $prednr = $1;
# my $signalnr = $2; # not used presently
if(! defined($gene)) {
# for GFF parser i modified the original GenScan module
# Magic Fang(fangl at genomics.org.cn) BGI, 2003-11-10
$gene = Bio::Tools::Prediction::Gene->new(-primary=>"gene",
'-display_name' => "GenePrediction$prednr",
'-source' => 'FgeneSH',
'-tag'=>{
'gene_id'=>"GenePrediction$prednr"});
}
# split into fields
chomp();
$_=~s/^\s+//g;
my @flds = split(/[\s|\t]+/, $_);
# print "@flds\n";
# create the feature object depending on the type of signal
my $predobj;
my $is_exon = grep {$_ eq $flds[3];} (keys(%exontags));
if($is_exon) {
$predobj = Bio::Tools::Prediction::Exon->new();
$predobj->score($flds[7]);
$predobj->start($flds[4]);
$predobj->end($flds[6]);
$predobj->start_signal_score($flds[7]);
$predobj->end_signal_score($flds[7]);
$predobj->coding_signal_score($flds[7]);
$predobj->significance($flds[7]);
# for correct GFF parser, i change the primary tag of the exon
# to the exon, and set the display name to the original exon
# primary tag
# Magic Fang(fangl at genomics.org.cn) BGI, 2003-11-10
$predobj->display_name($exontags{$flds[3]});
$predobj->primary_tag('exon');
$predobj->is_coding(1);
$predobj->seq_id($seqname);
# add 2 lines for artemis, for gene clustering and record the exon id
# in the last column of GFF
# Magic Fang(fangl at genomics.org.cn) BGI, 2003-11-10
$predobj->add_tag_value('exon_id', $exontags{$flds[3]}.'Exon');
$predobj->add_tag_value('gene', "GenePrediction$prednr");
# first, set fields unique to exons
# Figure out the frame of this exon. This is NOT the frame
# given by FgeneSH, which is the absolute frame of the base
# starting the first predicted complete codon. By comparing
# to the absolute frame of the first base we can compute the
# offset of the first complete codon to the first base of the
# exon, which determines the frame of the exon.
if($predobj->strand() == 1) {
$predobj->frame($flds[8]-$flds[4]);
} else {
$predobj->frame($flds[6]-$flds[10]);
}
# then add to gene structure object
# i modified the order of the follow 3 lines in GenScan module
# Magic Fang(fangl at genomics.org.cn) BGI, 2003-11-10
$predobj->source_tag('FgeneSH');
$predobj->strand((($flds[1] eq '+') ? 1 : -1));
$gene->add_exon($predobj, 'exon');
} else {
# PolyA site, or Promoter
$predobj = Bio::SeqFeature::Generic->new();
$predobj->score($flds[4]);
$predobj->start($flds[3]);
$predobj->end($flds[3]);
$predobj->source_tag('FgeneSH');
$predobj->strand((($flds[1] eq '+') ? 1 : -1));
$predobj->seq_id($seqname);
# add to gene structure (should be done only when start and end
# are set, in order to allow for proper expansion of the range)
if($flds[2] eq 'PolA') {
# for correct GFF and Feature Table format
# Magic Fang(fangl at genomics.org.cn) BGI, 2003-11-10
$predobj->primary_tag("PolyA");
$gene->poly_A_site($predobj);
} elsif($flds[2] eq 'TSS') {
$predobj->primary_tag("Promoter");
$gene->add_promoter($predobj);
}
next;
}
# set common fields
}
if(/^\s*$/ && defined($gene)) {
# current gene is completed
$gene->seq_id($seqname);
$self->_add_prediction($gene);
$gene = undef;
next;
}
if(/^(FGENESH)\s+(.+)\s+Prediction.+in\s+(\S+)\s+genomic/) {
$self->analysis_method($1);
$self->analysis_method_version($2);
$self->analysis_subject($3);
next;
}
if(/^\s*Seq name\:\s+(\S+)[\s|\r|\n]*/) {
$seqname = $1;
next;
}
$self->_has_cds(0);
/^>/ && do {
# section of predicted sequences
$self->_pushback($_);
last;
};
}
$self->_predictions_parsed(1);
}
=head2 _prediction
Title : _prediction()
Usage : $gene = $obj->_prediction()
Function: internal
Example :
Returns :
=cut
sub _prediction {
my ($self) = @_;
return undef unless(exists($self->{'_preds'}) && @{$self->{'_preds'}});
return shift(@{$self->{'_preds'}});
}
=head2 _add_prediction
Title : _add_prediction()
Usage : $obj->_add_prediction($gene)
Function: internal
Example :
Returns :
=cut
sub _add_prediction {
my ($self, $gene) = @_;
if(! exists($self->{'_preds'})) {
$self->{'_preds'} = [];
}
push(@{$self->{'_preds'}}, $gene);
}
=head2 _predictions_parsed
Title : _predictions_parsed
Usage : $obj->_predictions_parsed
Function: internal
Example :
Returns : TRUE or FALSE
=cut
sub _predictions_parsed {
my ($self, $val) = @_;
$self->{'_preds_parsed'} = $val if $val;
if(! exists($self->{'_preds_parsed'})) {
$self->{'_preds_parsed'} = 0;
}
return $self->{'_preds_parsed'};
}
=head2 _has_cds
Title : _has_cds()
Usage : $obj->_has_cds()
Function: Whether or not the result contains the predicted CDSs, too.
Example :
Returns : TRUE or FALSE
=cut
sub _has_cds {
my ($self, $val) = @_;
$self->{'_has_cds'} = $val if $val;
if(! exists($self->{'_has_cds'})) {
$self->{'_has_cds'} = 0;
}
return $self->{'_has_cds'};
}
=head2 _read_fasta_seq
Title : _read_fasta_seq()
Usage : ($id,$seqstr) = $obj->_read_fasta_seq();
Function: Simple but specialised FASTA format sequence reader. Uses
$self->_readline() to retrieve input, and is able to strip off
the traling description lines.
Example :
Returns : An array of two elements.
=cut
sub _read_fasta_seq {
my ($self) = @_;
my ($id, $seq);
local $/ = ">";
my $entry = $self->_readline();
if($entry) {
$entry =~ s/^>//;
# complete the entry if the first line came from a pushback buffer
while($entry !~ />$/) {
last unless $_ = $self->_readline();
$entry .= $_;
}
# delete everything onwards from an intervening empty line (at the
# end there might be statistics stuff)
$entry =~ s/[\n|\r]+>$//s;
# id and sequence
if($entry =~ /^(.+)\n([^>]+)/) {
$id = $1;
$seq = $2;
} else {
# I have to repair some fgenesh err here, for when the exon length is 3,
# fgenesh will not print the protein sequence, but just the fasta head,
# so i get rid of the exception handler here
# $self->throw("Can't parse FgeneSH predicted sequence entry");
# Magic Fang(fangl at genomics.org.cn) BGI, 2003-11-10
$seq=undef;
}
$seq =~ s/\s//g; # Remove whitespace
}
return ($id, $seq);
}
1;
More information about the Bioperl-l
mailing list