Bio/Tools/Signalp.pm

   1 # Parser module for Signalp Bio::Tools::Signalp
   2 #
   3 #
   4 # Based on the EnsEMBL module
   5 # Bio::EnsEMBL::Pipeline::Runnable::Protein::Signalp originally
   6 # written by Marc Sohrmann (ms2@sanger.ac.uk) Written in BioPipe by
   7 # Please direct questions and support issues to <bioperl-l@bioperl.org>
   8 #
   9 # Balamurugan Kumarasamy <savikalpa@fugu-sg.org> Cared for by the Fugu
  10 # Informatics team (fuguteam@fugu-sg.org)
  11
  12 # You may distribute this module under the same terms as perl itself
  13 #
  14 # POD documentation - main docs before the code
  15
  16 =head1 NAME
  17
  18 Bio::Tools::Signalp - parser for Signalp output
  19
  20 =head1 SYNOPSIS
  21
  22  use Bio::Tools::Signalp;
  23
  24  my $parser = Bio::Tools::Signalp->new(-fh =>$filehandle );
  25
  26  while( my $sp_feat = $parser->next_result ) {
  27    if ($sp_feat->score > 0.9) {
  28       push @likely_sigpep, $sp_feat;
  29    }
  30  }
  31
  32 =head1 DESCRIPTION
  33
  34 C<SignalP> predicts the presence and location of signal peptide
  35 cleavage sites in amino acid sequences.
  36
  37 L<Bio::Tools::Signalp> parses the output of C<SignalP> to provide a
  38 L<Bio::SeqFeature::Generic> object describing the signal peptide
  39 found, if any. It returns a variety of tags extracted from the NN and HMM
  40 analysis. Most importantly, the C<score()> attribute contains the
  41 NN probability of this being a true signal peptide.
  42
  43
  44 =head1 FEEDBACK
  45
  46 =head2 Mailing Lists
  47
  48  User feedback is an integral part of the evolution of this and other
  49  Bioperl modules. Send your comments and suggestions preferably to
  50  the Bioperl mailing list.  Your participation is much appreciated.
  51
  52   bioperl-l@bioperl.org                  - General discussion
  53   http://bioperl.org/wiki/Mailing_lists  - About the mailing lists
  54
  55 =head2 Support
  56
  57 Please direct usage questions or support issues to the mailing list:
  58
  59 I<bioperl-l@bioperl.org>
  60
  61 rather than to the module maintainer directly. Many experienced and
  62 reponsive experts will be able look at the problem and quickly
  63 address it. Please include a thorough description of the problem
  64 with code and data examples if at all possible.
  65
  66 =head2 Reporting Bugs
  67
  68 Report bugs to the Bioperl bug tracking system to help us keep track
  69 of the bugs and their resolution. Bug reports can be submitted va the
  70 web:
  71
  72   https://github.com/bioperl/bioperl-live/issues
  73
  74 =head1 AUTHOR
  75
  76 # Please direct questions and support issues to I<bioperl-l@bioperl.org>
  77
  78 Based on the EnsEMBL module Bio::EnsEMBL::Pipeline::Runnable::Protein::Signalp
  79 originally written by Marc Sohrmann (ms2_AT_sanger.ac.uk). Written in BioPipe by
  80 Balamurugan Kumarasamy savikalpa_AT_fugu-sg.org. Cared for by the Fugu
  81 Informatics team (fuguteam_AT_fugu-sg.org)
  82
  83 =head1 CONTRIBUTORS
  84
  85 Torsten Seemann - torsten.seemann AT infotech.monash.edu.au
  86
  87 =head1 APPENDIX
  88
  89 The rest of the documentation details each of the object methods.
  90 Internal methods are usually preceded with a _
  91
  92 =cut
  93
  94 package Bio::Tools::Signalp;
  95 use strict;
  96
  97 use Bio::SeqFeature::Generic;
  98 use base qw(Bio::Root::Root Bio::Root::IO);
  99
 100
 101
 102 =head2 new
 103
 104  Title   : new
 105  Usage   : my $obj = Bio::Tools::Signalp->new();
 106  Function: Builds a new Bio::Tools::Signalp object
 107  Returns : Bio::Tools::Signalp
 108  Args    : -fh/-file => $val, # for initing input, see Bio::Root::IO
 109
 110 =cut
 111
 112 sub new {
 113       my($class,@args) = @_;
 114
 115       my $self = $class->SUPER::new(@args);
 116       $self->_initialize_io(@args);
 117
 118       return $self;
 119 }
 120
 121 =head2 next_result
 122
 123  Title   : next_result
 124  Usage   : my $feat = $signalp->next_result
 125  Function: Get the next result set from parser data
 126  Returns : Bio::SeqFeature::Generic
 127  Args    : none
 128
 129 =cut
 130
 131 sub next_result {
 132         my ($self) = @_;
 133
 134         while (my $line=$self->_readline()) {
 135            chomp $line;
 136
 137            if ($line=~/^\>(\S+)/) {
 138               $self->_seqname($1);
 139            }
 140            elsif ($line=~/max\.\s+Y\s+(\S+)\s+\S+\s+\S+\s+(\S+)/) {
 141               $self->_fact1($2);
 142            }
 143            elsif ($line=~/mean\s+S\s+(\S+)\s+\S+\s+\S+\s+(\S+)/) {
 144               my $fact2 = $2;
 145
 146               if ($fact2 eq 'YES' and $self->_fact1 eq 'YES') {
 147
 148                   my $line = $self->_readline();
 149
 150                   ###########################################
 151                   # modification to suit new SignalP output
 152                   ###########################################
 153                                         chomp $line;
 154                                         #print STDERR "********** <$line>\n";
 155                                         if ($line =~ /\s+D\s+.*/) {
 156                                                 $line = $self->_readline();
 157                                         }
 158                                         #print STDERR "********** <$line>\n";
 159                                         my $end;
 160                                   ###########################################
 161
 162
 163                   if ($line =~ /Most likely cleavage site between pos\.\s+(\d+)/) {
 164                       my $end = $1;
 165                       my (%feature);
 166                       $feature{seq_id} = $self->_seqname;
 167                       $feature{start} = 1;
 168                       $feature{end} = $end;
 169                       $feature{source_tag} = 'Signalp';
 170                       $feature{primary}= 'signal_peptide';
 171                       $self->_parse_hmm_result(\%feature);
 172                       my $new_feat = $self->_create_feature (\%feature);
 173                       return $new_feat;
 174                   }
 175                   else {
 176                       $self->throw ("parsing problem in signalp");
 177                   }
 178
 179               }
 180            }
 181
 182         }
 183 }
 184
 185 =head2 _parse_hmm_result
 186
 187  Title   : _parse_hmm_result
 188  Usage   : $self->_parse_hmm_result(\%feature)
 189  Function: Internal (not to be used directly)
 190  Returns : hash of feature values
 191  Args    : hash of more feature values
 192
 193 =cut
 194
 195 sub _parse_hmm_result {
 196     my ($self, $feature_hash) = @_;
 197     while(my $line = $self->_readline){
 198         chomp $line;
 199         if($line =~ /Prediction: (.+)$/){
 200             $feature_hash->{hmmProdiction} = $1;
 201         }elsif($line =~ /Signal peptide probability: ([0-9\.]+)/){
 202             $feature_hash->{peptideProb} = $1;
 203         }elsif($line =~ /Signal anchor probability: ([0-9\.]+)/){
 204             $feature_hash->{anchorProb} = $1;
 205             last;
 206         }
 207     }
 208 }
 209
 210 =head2 _create_feature
 211
 212  Title   : _create_feature
 213  Usage   : $self->create_feature(\%feature)
 214  Function: Internal (not to be used directly)
 215  Returns : hash of feature values
 216  Args    : hash of more feature values
 217
 218 =cut
 219
 220 sub _create_feature {
 221     my ($self, $feat) = @_;
 222
 223     # create feature object
 224     my $feature = Bio::SeqFeature::Generic->new(
 225          -seq_id      => $feat->{name},
 226          -start       => $feat->{start},
 227          -end         => $feat->{end},
 228          -score       => $feat->{score},
 229          -source      => $feat->{source},
 230          -primary     => $feat->{primary},
 231          -logic_name  => $feat->{logic_name},
 232     );
 233
 234     $feature->score($feat->{peptideProb});
 235     $feature->add_tag_value('peptideProb', $feat->{peptideProb});
 236     $feature->add_tag_value('anchorProb', $feat->{anchorProb});
 237     $feature->add_tag_value('evalue',$feat->{anchorProb});
 238     $feature->add_tag_value('percent_id','NULL');
 239     $feature->add_tag_value("hid",$feat->{primary});
 240     $feature->add_tag_value('SignalpPrediction', $feat->{hmmProdiction});
 241     return $feature;
 242
 243 }
 244
 245 =head2 _seqname
 246
 247  Title   : _seqname
 248  Usage   : $self->_seqname($name)
 249  Function: Internal (not to be used directly)
 250  Returns :
 251  Args    :
 252
 253 =cut
 254
 255 sub _seqname{
 256     my ($self,$seqname)=@_;
 257
 258     if (defined$seqname){
 259         $self->{'seqname'}=$seqname;
 260     }
 261     return $self->{'seqname'};
 262 }
 263
 264 =head2 _fact1
 265
 266  Title   : _fact1
 267  Usage   : $self->fact1($fact1)
 268  Function: Internal (not to be used directly)
 269  Returns :
 270  Args    :
 271
 272 =cut
 273
 274 sub _fact1{
 275     my ($self, $fact1)=@_;
 276
 277     if (defined $fact1){
 278        $self->{'fact1'}=$fact1;
 279     }
 280     return $self->{'fact1'};
 281 }
 282
 283
 284
 285 1;
 286
 287