Bio/Tools/TargetP.pm

   1 #
   2 # Bioperl module for TargetP
   3 #
   4 # Please direct questions and support issues to <bioperl-l@bioperl.org>
   5 #
   6 # Cared for by Emmanuel Quevillon <emmanuel.quevillon@versailles.inra.fr>
   7 #
   8 # You may distribute this module under the same terms as perl itself
   9 #
  10 # POD documentation - main docs before the code
  11
  12 =head1 NAME
  13
  14 Bio::Tools::TargetP - Results of one TargetP run
  15
  16 =head1 SYNOPSIS
  17
  18    use Bio::Tools::TargetP;
  19
  20    #filename for  TargetP result :
  21    $targetp = Bio::Tools::TargetP->new(-file => 'targetp.out');
  22
  23    # filehandle for TargetP :
  24    $targetp = Bio::Tools::TargetP->new( -fh  => \*INPUT );
  25
  26    ### targetp v1.1 prediction results ##################################
  27    #Number of query sequences:  11
  28    #Cleavage site predictions included.
  29    #Using NON-PLANT networks.
  30    #
  31    #Name                  Len            mTP     SP  other  Loc  RC  TPlen
  32    #----------------------------------------------------------------------
  33    #swall|Q9LIP3|C72Y_AR  500          0.245  0.935  0.009   S    2     22
  34    #swall|Q52813|AAPQ_RH  400          0.170  0.462  0.577   _    5      -
  35    #swall|O86459|AAT_RHI  400          0.346  0.046  0.660   _    4      -
  36
  37
  38
  39    # parse the results
  40    while($feature = $targetp->next_prediction()) {
  41
  42            #$feature is a Bio::SeqFeature::Generic object
  43            my $method     = $targetp->analysis_method();
  44            my $vesion     = $targetp->analysis_method_version() || $feature->source();
  45            my $seqid      = $feature->seq_id();
  46            # ...
  47      }
  48
  49    # essential if you gave a filename at initialization (otherwise the file
  50    # will stay open)
  51    $targetp->close();
  52
  53 =head1 DESCRIPTION
  54
  55 TargetP modules will provides parsed information about protein
  56 localization.  It reads in a targetp output file.  It parses the
  57 results, and returns a Bio::SeqFeature::Generic object for each
  58 seqeunces found to have a subcellular localization
  59
  60 =head1 FEEDBACK
  61
  62 =head2 Mailing Lists
  63
  64 User feedback is an integral part of the evolution of this and other
  65 Bioperl modules. Send your comments and suggestions preferably to
  66 the Bioperl mailing list.  Your participation is much appreciated.
  67
  68   bioperl-l@bioperl.org                  - General discussion
  69   http://bioperl.org/wiki/Mailing_lists  - About the mailing lists
  70
  71 =head2 Support
  72
  73 Please direct usage questions or support issues to the mailing list:
  74
  75 I<bioperl-l@bioperl.org>
  76
  77 rather than to the module maintainer directly. Many experienced and
  78 reponsive experts will be able look at the problem and quickly
  79 address it. Please include a thorough description of the problem
  80 with code and data examples if at all possible.
  81
  82 =head2 Reporting Bugs
  83
  84 Report bugs to the Bioperl bug tracking system to help us keep track
  85 of the bugs and their resolution. Bug reports can be submitted via
  86 the web:
  87
  88   https://redmine.open-bio.org/projects/bioperl/
  89
  90 =head1 AUTHORS - Emmanuel Quevillon
  91
  92 Email emmanuel.quevillon@versailles.inra.fr
  93
  94 Describe contact details here
  95
  96 =head1 APPENDIX
  97
  98 The rest of the documentation details each of the object methods.
  99 Internal methods are usually preceded with a _
 100
 101 =cut
 102
 103
 104 # Let the code begin...
 105
 106
 107 package Bio::Tools::TargetP;
 108 use strict;
 109 use Bio::Tools::AnalysisResult;
 110 use Bio::SeqFeature::Generic;
 111 use Data::Dumper;
 112
 113 use base qw(Bio::Tools::AnalysisResult);
 114
 115
 116 #Definition of 'Loc' field according to http://www.cbs.dtu.dk/services/TargetP/output.php
 117 my $MAPLOC = {
 118               'S' => 'Secretory pathway',
 119               'M' => 'Mitochondrion',
 120               'C' => 'Chloroplast',
 121               '_' => 'Any other',
 122               '*' => 'Unknown',
 123               '?' => 'Unknown',
 124              };
 125
 126
 127 =head1 analysis_method
 128
 129  Usage     : $self->analysis_method();
 130  Purpose   : Inherited method. Overridden to ensure that the name matches
 131  Returns   : String
 132  Argument  : n/a
 133
 134 =cut
 135
 136 sub analysis_method {
 137
 138     my ($self, $method) = @_;
 139
 140     if($method && ($method !~ /TargetP/i)) {
 141         $self->throw("method $method not supported in " . ref($self));
 142     }
 143
 144     return $self->SUPER::analysis_method($method);
 145 }
 146
 147 =head1 network
 148
 149   Title   : network
 150   Usage   : $self->network($network)
 151   Function: This method Get/Set the network used for the analysis (PLANT or NON-PLANT)
 152   Example :
 153   Returns : string
 154   Arguments: On set, the network used
 155
 156 =cut
 157
 158 sub network {
 159
 160     my($self, $net) = @_;
 161
 162     if(defined($net)){
 163         $self->{'_network'} = $net;
 164     }
 165
 166     return $self->{'_network'};
 167
 168 }
 169
 170
 171 =head1 cleavage
 172
 173   Title    :  cleavage
 174   Usage    : $self->cleavage($cleavage)
 175   Function : This method Get/Set if SignalP program was used to run TargetP
 176   Example  :
 177   Returns  : 1 or 0
 178   Arguments: On set, the cleavage used or not
 179
 180 =cut
 181
 182 sub cleavage {
 183
 184     my($self, $cleavage) = @_;
 185
 186     if(defined($cleavage)){
 187         $self->{'_cleavage'} = $cleavage =~ /not included/ ? '0' : '1';
 188     }
 189
 190     return $self->{'_cleavage'};
 191
 192 }
 193
 194
 195 =head1 next_prediction
 196
 197   Usage    : $targetp->next_prediction()
 198   Purpose  : Returns the next TargetP prediction
 199   Returns  : A Bio::SeqFeature::Generic object
 200   Arguments: n/a
 201
 202 =cut
 203
 204 sub next_prediction {
 205
 206     my($self) = @_;
 207
 208     unless($self->_parsed()){
 209         $self->_parse_results();
 210         $self->_parsed(1);
 211     }
 212
 213     return shift @{$self->{'_features'}} || undef;
 214 }
 215
 216 =head1 create_feature
 217
 218   Title     : create_feature
 219   Usage     : $self->create_feature(\%hash);
 220   Function  : This method creates a new Bio::SeqFeature::Generic object
 221   Example   :
 222   Returns   : Bio::SeqFeature::Generic
 223   Arguments : hash reference
 224
 225 =cut
 226
 227 sub create_feature {
 228
 229     my($self, $feat) = @_;
 230
 231     $self->throw("Need a reference to hash table") unless($feat && ref($feat) eq 'HASH');
 232
 233     my $feature = Bio::SeqFeature::Generic->new(
 234                                                 -seq_id      => $feat->{seqid},
 235                                                 -source_tag  => $self->analysis_method(),
 236                                                 -primary_tag => 'signal_peptide',        #Sequence Ontology compliant
 237                                                 -strand      => '+',
 238                                                );
 239
 240     if(defined($feat->{seqlen})){
 241         $feature->start(1);
 242         $feature->end($feat->{seqlen});
 243     }
 244     $feature->add_tag_value('location',            $MAPLOC->{$feat->{loc}})   if(exists($MAPLOC->{$feat->{loc}}));
 245     $feature->add_tag_value('chloroplastCutOff',   $feat->{cTP})              if(defined($feat->{cTP}));
 246     $feature->add_tag_value('mitochondrionCutOff', $feat->{mTP})              if(defined($feat->{mTP}));
 247     $feature->add_tag_value('signalPeptideCutOff', $feat->{SP})               if(defined($feat->{SP}));
 248     $feature->add_tag_value('otherCutOff',         $feat->{other})            if(defined($feat->{other}));
 249     $feature->add_tag_value('reliabilityClass',    $feat->{RC})               if(defined($feat->{RC}));
 250     $feature->add_tag_value('signalPeptideLength', $feat->{TPLen})            if(defined($feat->{TPLen}));
 251
 252     $feature->add_tag_value('network',             $self->network());
 253
 254     return $feature;
 255
 256 }
 257
 258
 259 =head2 PRIVATE METHODS
 260
 261 =cut
 262
 263 =head2 _initialize_state
 264
 265  Title   : _initialize_state
 266  Usage   : n/a; usually called by _initialize() itself called by new()
 267  Function: This method is supposed to reset the state such that any 'history'
 268            is lost. State information that does not change during object
 269            lifetime is not considered as history, e.g. parent, name, etc shall
 270            not be reset. An inheriting object should only be concerned with
 271            state information it introduces itself, and for everything else
 272            call SUPER::_initialize_state(@args).
 273
 274            The argument syntax is the same as for new() and _initialize(),
 275            i.e., named parameters following the -name=>$value convention.
 276            The following parameters are dealt with by the implementation
 277            provided here:
 278               -INPUT, -FH, -FILE
 279            (tags are case-insensitive).
 280  Example :
 281  Returns :
 282  Args    :
 283
 284 =cut
 285
 286 sub _initialize_state  {
 287
 288         my ($self,@args,) = @_;
 289         # first call the inherited method!
 290         $self->SUPER::_initialize_state(@args);
 291
 292         # our private state variables
 293         $self->{'_features'}   = [ ];
 294         $self->{'_parameters'} = undef;
 295         $self->{'_format'}     = undef;
 296         $self->{'_network'}    = undef;
 297         $self->{'_cleavage'}   = undef;
 298         $self->{'_parsed'}     = 0;
 299
 300         $self->analysis_method('TargetP');
 301
 302         return 1;
 303 }
 304
 305 =head2 _predictions
 306
 307   Usage    : $targetp->_prediction()
 308   Purpose  : Returns the number of TargetP predictions
 309   Returns  : A scalar (number)
 310   Arguments: n/a
 311
 312 =cut
 313
 314 sub _predictions {
 315
 316     my($self) = @_;
 317
 318     return scalar(@{$self->{'_features'}}) || 0;
 319 }
 320
 321
 322 =head2 _parsed
 323
 324  Title     : _parsed
 325  Usage     : $targetp->_parsed(1)
 326  Function  : This method is used to know if the output result is parsed or not
 327              For internal use only
 328  Example   :
 329  Returns   : 1/0
 330  Arguments : 1/0 for setting
 331
 332 =cut
 333
 334 sub _parsed {
 335
 336     my($self, $value) = @_;
 337
 338     if(defined($value)){
 339         $self->{'_parsed'} = $value;
 340     }
 341
 342     return $self->{'_parsed'};
 343 }
 344
 345
 346
 347 =head2 _parse_results
 348
 349   Title    : _parse_results
 350   Usage    : $self->_parse_results()
 351   Function : This method parses a TargetP output
 352              For internal use only
 353   Example  :
 354   Returns  : n/a
 355   Arguments: none
 356
 357 =cut
 358
 359 sub _parse_results {
 360
 361     my($self) = @_;
 362
 363
 364     ### targetp v1.1 prediction results ##################################
 365     #Number of query sequences:  11
 366     #Cleavage site predictions included.
 367     #Using NON-PLANT networks.
 368     #
 369     #Name                  Len            mTP     SP  other  Loc  RC  TPlen
 370     #----------------------------------------------------------------------
 371     #swall|Q9LIP3|C72Y_AR  500          0.245  0.935  0.009   S    2     22
 372     #swall|Q52813|AAPQ_RH  400          0.170  0.462  0.577   _    5      -
 373     #swall|O86459|AAT_RHI  400          0.346  0.046  0.660   _    4      -
 374
 375
 376     while(defined(my $line = $self->_readline())){
 377
 378         if($line =~ /targetp (v[\d\.]+)/){
 379
 380             $self->analysis_method_version($1);
 381
 382         }elsif($line =~ /Cleavage site predictions (.*)/){
 383
 384             $self->cleavage($1);
 385
 386         }elsif($line =~ /Using (\S+) networks/){
 387
 388             $self->network($1);
 389
 390         }elsif($line =~ /^Name/){
 391
 392             #We skip the next line which is '------------------'
 393             $self->_readline();
 394
 395             my $hash = { };
 396
 397             while(defined(my $line = $self->_readline())){
 398
 399                 last if($line =~ /^----/);
 400
 401                 my $hash = $self->_parse_line($line);
 402
 403                 my $new_feature = $self->create_feature($hash);
 404
 405                 $self->_add_feature($new_feature);
 406             }
 407         }
 408     }
 409
 410     return;
 411 }
 412
 413 =head2 _parse_line
 414
 415  Title    : _parse_line
 416  Usage    : $self->_parse_line($line)
 417  Function : This method parses the line result
 418             For internal use only
 419  Example  :
 420  Returns  : Hash reference
 421  Arguemnts: line to parse
 422
 423 =cut
 424
 425 sub _parse_line {
 426
 427     my($self, $line) = @_;
 428
 429     $self->throw("No line to parse given") unless($line);
 430
 431     my $hash = { };
 432     my ($seqid, $seqlen, $cTP, $mTP, $SP, $other, $loc, $RC, $TPlen);
 433
 434     if($self->network() eq 'NON-PLANT'){
 435
 436         ($seqid, $seqlen, $mTP, $SP, $other, $loc, $RC, $TPlen) = split(/\s+/, $line);
 437
 438     }else{
 439
 440         ($seqid, $seqlen, $cTP, $mTP, $SP, $other, $loc, $RC, $TPlen) = split(/\s+/, $line);
 441
 442     }
 443
 444     $hash->{seqid}  = $seqid;
 445     $hash->{seqlen} = $seqlen;
 446     $hash->{cTP}    = $cTP || undef;
 447     $hash->{mTP}    = $mTP;
 448     $hash->{SP}     = $SP;
 449     $hash->{other}  = $other;
 450     $hash->{loc}    = $loc;
 451     $hash->{RC}     = $RC;
 452     $hash->{TPLen}  = ($TPlen && $TPlen =~ /\d+/) ? $TPlen : undef;
 453
 454     return $hash;
 455
 456 }
 457
 458 =head2 _add_feature
 459
 460  Title    : _add_feature
 461  Usage    : $self->_add_feature($feature)
 462  Function : This method stores a feature object
 463             For internal use only
 464  Example  :
 465  Returns  : n/a
 466  Arguments: Bio::SeqFeature::Generic
 467
 468 =cut
 469
 470 sub _add_feature {
 471
 472     my($self, $feature) = @_;
 473
 474     $self->throw("Need a Bio::SeqFeature::Generic object") unless $feature->isa("Bio::SeqFeature::Generic");
 475
 476     push(@{$self->{'_features'}}, $feature);
 477
 478     return;
 479
 480 }
 481
 482 =head2 _toString_location
 483
 484  Title    : _toString_location
 485  Usage    : $self->_toString_location($key)
 486  Function : This method convert the 'one letter code' location to
 487             the corresponding definition
 488             For internal use only
 489  Example  :
 490  Returns  : Location or undef
 491  Arguments: String
 492
 493 =cut
 494
 495 sub _toString_location {
 496
 497     my($self, $key) = @_;
 498
 499     if($key && exists($MAPLOC->{$key})){
 500         return $MAPLOC->{$key};
 501     }
 502
 503     return;
 504 }
 505
 506
 507
 508 1;