t/Tools/Signalp/ExtendedSignalp.t

   1 # -*-Perl-*- Test Harness script for Bioperl
   2 # $Id$
   3
   4 use strict;
   5 use Data::Dumper;
   6 BEGIN {
   7     use Bio::Root::Test;
   8
   9     test_begin(-tests => 185);
  10
  11     use_ok('Bio::Tools::Signalp::ExtendedSignalp');
  12 }
  13
  14 ###############################################
  15 ### TESTS ON SUMMARY OUTPUT FORMAT (NN+HMM) ###
  16 ###############################################
  17
  18 my $res = {
  19            '1' => {
  20                    'id'     => 'BC1G_00003.1',
  21                    'pred'   => 'Signal peptide',
  22                    'nnpred' => 'signal-peptide',
  23                    'end'    => '22',
  24                    'prob'   => '0.999',
  25                    'anchor' => '0.000',
  26                   },
  27         '2' => {
  28                    'id'     => 'BC1G_00008.1',
  29                    'pred'   => 'Non-secretory protein',
  30                    'nnpred' => 'signal-peptide',
  31                    'end'    => '83',
  32                    'prob'   => '0.222',
  33                    'anchor' => '0.067',
  34                   },
  35           };
  36
  37 # Test on filtered results
  38 my $facts   = [qw(maxS D)];
  39 my $in      = test_input_file("signalp.summary");
  40 my $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
  41                                                         -file => $in,
  42                                                         -factors => $facts,
  43                                                        );
  44
  45 ok($signalp);
  46 my $i = 1;
  47
  48 while(my $feat = $signalp->next_feature()){
  49     #print Dumper($feat);
  50     is($feat->seq_id(), $res->{$i}->{id});
  51     is($feat->end(), $res->{$i}->{end});
  52     is(($feat->get_tag_values('peptideProb'))[0], $res->{$i}->{prob});
  53     is(($feat->get_tag_values('anchorProb'))[0], $res->{$i}->{anchor});
  54     is(($feat->get_tag_values('signalpPrediction'))[0], $res->{$i}->{pred});
  55     is(($feat->get_tag_values('nnPrediction'))[0], $res->{$i}->{nnpred});
  56     $i++;
  57 }
  58
  59 # Tests without filters.
  60 # It should by default only parses results with Ymax and meanS to mimic default behavior
  61 # from Bio::Tools::Signalp
  62 $res = {
  63         '1' => {
  64                 'id'     => 'BC1G_00003.1',
  65                 'pred'   => 'Signal peptide',
  66                 'end'    => '22',
  67                 'prob'   => '0.999',
  68                 'anchor' => '0.000',
  69                },
  70         '2' => {
  71                 'id'     => 'BC1G_00008.1',
  72                 'pred'   => 'Non-secretory protein',
  73                 'end'    => '83',
  74                 'prob'   => '0.222',
  75                 'anchor' => '0.067',
  76                },
  77        };
  78
  79 #No filters required
  80 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
  81                                                      -file => $in
  82                                                     );
  83 ok($signalp);
  84 $i = 1;
  85
  86 while(my $feat = $signalp->next_feature()){
  87     is($feat->seq_id(), $res->{$i}->{id});
  88     is($feat->end(), $res->{$i}->{end});
  89     is(($feat->get_tag_values('peptideProb'))[0], $res->{$i}->{prob});
  90     is(($feat->get_tag_values('anchorProb'))[0], $res->{$i}->{anchor});
  91     is(($feat->get_tag_values('signalpPrediction'))[0], $res->{$i}->{pred});
  92     $i++;
  93 }
  94
  95 #############################################
  96 ### TESTS ON SHORT OUTPUT FORMAT (NN+HMM) ###
  97 #############################################
  98
  99 $res = {
 100         '1' => {
 101                 'id'     => 'BC1G_00003.1',
 102                 'pred'   => 'Signal peptide',
 103                 'nnpred' => 'signal-peptide',
 104                 'end'    => '22',
 105                },
 106         '2' => {
 107                 'id'     => 'BC1G_00008.1',
 108                 'nnpred' => 'signal-peptide',
 109                 'pred'   => 'Non-secretory protein',
 110                 'end'    => '83',
 111                },
 112         '3' => {
 113                 'id'     => 'BC1G_00009.1',
 114                 'nnpred' => 'signal-peptide',
 115                 'pred'   => 'Non-secretory protein',
 116                 'end'    => '28',
 117                 },
 118         '4' => {
 119                 'id'     => 'BC1G_00010.1',
 120                 'nnpred' => 'signal-peptide',
 121                 'pred'   => 'Non-secretory protein',
 122                 'end'    => '15',
 123                 },
 124
 125        };
 126
 127 # Test on filtered results
 128 $facts   = [qw(maxC)];
 129 $in      = test_input_file("signalp.short");
 130 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
 131                                                      -file => $in,
 132                                                      -factors => $facts,
 133                                                     );
 134
 135 ok($signalp);
 136 $i = 1;
 137
 138 while(my $feat = $signalp->next_feature()){
 139     is($feat->seq_id(), $res->{$i}->{id});
 140     is($feat->end(), $res->{$i}->{end});
 141     is(($feat->get_tag_values('signalpPrediction'))[0], $res->{$i}->{pred});
 142     is(($feat->get_tag_values('nnPrediction'))[0], $res->{$i}->{nnpred});
 143     $i++;
 144 }
 145
 146 # Tests without filters.
 147 # It should by default only parses results with Ymax and meanS to mimic default behavior
 148 # from GPI::Bio::Tools::Signalp
 149 $res = {
 150         '1' => {
 151                 'id'     => 'BC1G_00003.1',
 152                 'pred'   => 'Signal peptide',
 153         'prob' => 0.999,
 154                 'nnpred' => 'signal-peptide',
 155                 'end'    => '22',
 156                },
 157         '2' => {
 158                 'id'     => 'BC1G_00008.1',
 159                 'pred'   => 'Non-secretory protein',
 160         'prob' => 0.222,
 161                 'nnpred' => 'signal-peptide',
 162                 'end'    => '83',
 163                },
 164        };
 165
 166 #No filters required
 167 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
 168                                                      -file => $in
 169                                                     );
 170 ok($signalp);
 171 $i = 1;
 172
 173 while(my $feat = $signalp->next_feature()){
 174     is($feat->seq_id(), $res->{$i}->{id});
 175     is($feat->end(), $res->{$i}->{end});
 176     is(($feat->get_tag_values('peptideProb'))[0], $res->{$i}->{prob});
 177     is(($feat->get_tag_values('anchorProb'))[0], $res->{$i}->{anchor});
 178     is(($feat->get_tag_values('signalpPrediction'))[0], $res->{$i}->{pred});
 179     is(($feat->get_tag_values('nnPrediction'))[0], $res->{$i}->{nnpred});
 180     $i++;
 181 }
 182
 183
 184 ###########################################
 185 ### TESTS ON SUMMARY OUTPUT FORMAT (NN) ###
 186 ###########################################
 187
 188 $res = {
 189         '1' => {
 190                 'id'     => 'BC1G_00003.1',
 191                 'nnpred' => 'signal-peptide',
 192                 'end'    => '22',
 193                },
 194         '2' => {
 195                 'id'     => 'BC1G_00008.1',
 196                 'nnpred' => 'signal-peptide',
 197                 'end'    => '83',
 198                },
 199        };
 200
 201 # Test on filtered results BROKEN
 202 $facts   = [qw(maxC)];
 203 $in      = test_input_file("signalp.nn.summary");
 204 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
 205                                                      -file => $in,
 206                                                      -factors => $facts,
 207                                                     );
 208
 209 ok($signalp);
 210 $i = 1;
 211
 212 while(my $feat = $signalp->next_feature()){
 213
 214     is($feat->seq_id(), $res->{$i}->{id});
 215     is($feat->end(), $res->{$i}->{end});
 216     is(($feat->get_tag_values('nnPrediction'))[0], $res->{$i}->{nnpred});
 217     $i++;
 218 }
 219 # Tests without filters.
 220 # It should by default only parses results with Ymax and meanS to mimic default behavior
 221 # from GPI::Bio::Tools::Signalp
 222 $res = {
 223         '1' => {
 224                 'id'     => 'BC1G_00003.1',
 225                 'nnpred' => 'signal-peptide',
 226                 'end'    => '22',
 227                },
 228         '2' => {
 229                 'id'     => 'BC1G_00008.1',
 230                 'nnpred' => 'signal-peptide',
 231                 'end'    => '83',
 232                },
 233         '3' => {
 234                 'id'     => 'BC1G_00009.1',
 235                 'nnpred' => 'signal-peptide',
 236                 'end'    => '28',
 237                 },
 238         '4' => {
 239                 'id'     => 'BC1G_00010.1',
 240                 'nnpred' => 'signal-peptide',
 241                 'end'    => '15',
 242                 },
 243        };
 244
 245 #No filters required BROKEN
 246 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
 247                                                      -file => $in
 248                                                     );
 249 ok($signalp);
 250 $i = 1;
 251
 252 while(my $feat = $signalp->next_feature()){
 253
 254     is($feat->seq_id(), $res->{$i}->{id});
 255     is($feat->end(), $res->{$i}->{end});
 256     is(($feat->get_tag_values('nnPrediction'))[0], $res->{$i}->{nnpred});
 257     $i++;
 258 }
 259
 260
 261 ############################################
 262 ### TESTS ON SUMMARY OUTPUT FORMAT (HMM) ###
 263 ############################################
 264
 265 $res = {
 266         '1' => {
 267                 'id'     => 'BC1G_00002.1',
 268                 'prob'   => '0.000',
 269                 'anchor' => '0.000',
 270                 'cleav'  => '0.000',
 271                 'pred'   => 'Non-secretory protein',
 272                 'end'    => '22',
 273                },
 274         '2' => {
 275                 'id'     => 'BC1G_00003.1',
 276                 'prob'   => '0.999',
 277                 'anchor' => '0.000',
 278                 'cleav'  => '0.973',
 279                 'pred'   => 'Signal peptide',
 280                 'end'    => '22',
 281                },
 282         '3' => {
 283                 'id'     => 'BC1G_00004.1',
 284                 'prob'   => '0.003',
 285                 'anchor' => '0.000',
 286                 'cleav'  => '0.001',
 287                 'pred'   => 'Non-secretory protein',
 288                 'end'    => '19',
 289                },
 290         '4' => {
 291                 'id'     => 'BC1G_00005.1',
 292                 'prob'   => '0.008',
 293                 'anchor' => '0.000',
 294                 'cleav'  => '0.007',
 295                 'pred'   => 'Non-secretory protein',
 296                 'end'    => '22',
 297                 },
 298         '5' => {
 299                 'id'     => 'BC1G_00006.1',
 300                 'prob'   => '0.000',
 301                 'anchor' => '0.000',
 302                 'cleav'  => '0.000',
 303                 'pred'   => 'Non-secretory protein',
 304                 'end'    => '23',
 305                 },
 306         '6' => {
 307                 'id'     => 'BC1G_00007.1',
 308                 'prob'   => '0.240',
 309                 'anchor' => '0.000',
 310                 'cleav'  => '0.228',
 311                 'pred'   => 'Non-secretory protein',
 312                 'end'    => '22',
 313                 },
 314         '7' => {
 315                 'id'     => 'BC1G_00008.1',
 316                 'prob'   => '0.222',
 317                 'anchor' => '0.067',
 318                 'cleav'  => '0.061',
 319                 'pred'   => 'Non-secretory protein',
 320                 'end'    => '22',
 321                 },
 322         '8' => {
 323                 'id'     => 'BC1G_00009.1',
 324                 'prob'   => '0.000',
 325                 'anchor' => '0.000',
 326                 'cleav'  => '0.000',
 327                 'pred'   => 'Non-secretory protein',
 328                 'end'    => '20',
 329                 },
 330        };
 331
 332 # It is impossible to filter with hmm output...
 333 $in      = test_input_file("signalp.hmm.summary");
 334 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
 335                                                      -file => $in,
 336                                                     );
 337
 338 ok($signalp);
 339 $i = 1;
 340
 341 while(my $feat = $signalp->next_feature()){
 342     is($feat->seq_id(), $res->{$i}->{id});
 343     is($feat->end(), $res->{$i}->{end});
 344     is(($feat->get_tag_values('peptideProb'))[0], $res->{$i}->{prob});
 345     is(($feat->get_tag_values('anchorProb'))[0], $res->{$i}->{anchor});
 346     is(($feat->get_tag_values('signalpPrediction'))[0], $res->{$i}->{pred});
 347     is(($feat->get_tag_values('cleavageSiteProb'))[0], $res->{$i}->{cleav});
 348     $i++;
 349 }
 350
 351 #########################################
 352 ### TESTS ON SHORT OUTPUT FORMAT (NN) ###
 353 #########################################
 354
 355 $res = {
 356         '1' => {
 357                 'id'     => 'BC1G_00003.1',
 358                 'nnpred' => 'signal-peptide',
 359                 'yprob'  => '0.866',
 360                 'dprob'  => '0.902',
 361                 'cprob'  => '0.934',
 362                 'end'    => '22',
 363                },
 364         '2' => {
 365                 'id'     => 'BC1G_00008.1',
 366                 'yprob'  => '0.383',
 367                 'dprob'  => '0.436',
 368                 'cprob'  => '0.576',
 369                 'nnpred' => 'signal-peptide',
 370                 'end'    => '83',
 371                },
 372        };
 373
 374 # Test on filtered results
 375 $facts   = [qw(maxY)];
 376 $in      = test_input_file("signalp.nn.short");
 377 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
 378                                                      -file => $in,
 379                                                      -factors => $facts,
 380                                                     );
 381
 382 ok($signalp);
 383 $i = 1;
 384
 385 while(my $feat = $signalp->next_feature()){
 386     is($feat->seq_id(), $res->{$i}->{id});
 387     is($feat->end(), $res->{$i}->{end});
 388     is(($feat->get_tag_values('maxCprob'))[0], $res->{$i}->{cprob});
 389     is(($feat->get_tag_values('Dprob'))[0], $res->{$i}->{dprob});
 390     is(($feat->get_tag_values('maxYprob'))[0], $res->{$i}->{yprob});
 391     is(($feat->get_tag_values('nnPrediction'))[0], $res->{$i}->{nnpred});
 392     $i++;
 393 }
 394 # Tests without filters.
 395 # It should by default only parses results with Ymax and meanS to mimic default behavior
 396 # from GPI::Bio::Tools::Signalp
 397 $res = {
 398         '1' => {
 399                 'id'     => 'BC1G_00003.1',
 400                 'nnpred' => 'signal-peptide',
 401                 'yprob'  => '0.866',
 402                 'dprob'  => '0.902',
 403                 'cprob'  => '0.934',
 404                 'end'    => '22',
 405                },
 406         '2' => {
 407                 'id'     => 'BC1G_00008.1',
 408                 'yprob'  => '0.383',
 409                 'dprob'  => '0.436',
 410                 'cprob'  => '0.576',
 411                 'nnpred' => 'signal-peptide',
 412                 'end'    => '83',
 413                },
 414        };
 415
 416 #No filters required
 417 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
 418                                                      -file => $in
 419                                                     );
 420 ok($signalp);
 421 $i = 1;
 422
 423 while(my $feat = $signalp->next_feature()){
 424     is($feat->seq_id(), $res->{$i}->{id});
 425     is($feat->end(), $res->{$i}->{end});
 426     is(($feat->get_tag_values('maxCprob'))[0], $res->{$i}->{cprob});
 427     is(($feat->get_tag_values('Dprob'))[0], $res->{$i}->{dprob});
 428     is(($feat->get_tag_values('maxYprob'))[0], $res->{$i}->{yprob});
 429     is(($feat->get_tag_values('nnPrediction'))[0], $res->{$i}->{nnpred});
 430     $i++;
 431 }
 432
 433 ##########################################
 434 ### TESTS ON SHORT OUTPUT FORMAT (HMM) ###
 435 ##########################################
 436
 437 $res = {
 438         '1' => {
 439                 'id'     => 'BC1G_00002.1',
 440                 'prob'   => '0.000',
 441                 'cleav'  => '0.000',
 442                 'pred'   => 'Non-secretory protein',
 443                 'end'    => '22',
 444                },
 445         '2' => {
 446                 'id'     => 'BC1G_00003.1',
 447                 'prob'   => '0.999',
 448                 'cleav'  => '0.973',
 449                 'pred'   => 'Signal peptide',
 450                 'end'    => '22',
 451                },
 452         '3' => {
 453                 'id'     => 'BC1G_00004.1',
 454                 'prob'   => '0.003',
 455                 'cleav'  => '0.001',
 456                 'pred'   => 'Non-secretory protein',
 457                 'end'    => '19',
 458                },
 459         '4' => {
 460                 'id'     => 'BC1G_00005.1',
 461                 'prob'   => '0.008',
 462                 'cleav'  => '0.007',
 463                 'pred'   => 'Non-secretory protein',
 464                 'end'    => '22',
 465                 },
 466         '5' => {
 467                 'id'     => 'BC1G_00006.1',
 468                 'prob'   => '0.000',
 469                 'cleav'  => '0.000',
 470                 'pred'   => 'Non-secretory protein',
 471                 'end'    => '23',
 472                 },
 473         '6' => {
 474                 'id'     => 'BC1G_00007.1',
 475                 'prob'   => '0.240',
 476                 'cleav'  => '0.228',
 477                 'pred'   => 'Non-secretory protein',
 478                 'end'    => '22',
 479                 },
 480         '7' => {
 481                 'id'     => 'BC1G_00008.1',
 482                 'prob'   => '0.222',
 483                 'cleav'  => '0.061',
 484                 'pred'   => 'Non-secretory protein',
 485                 'end'    => '22',
 486                 },
 487         '8' => {
 488                 'id'     => 'BC1G_00009.1',
 489                 'prob'   => '0.000',
 490                 'cleav'  => '0.000',
 491                 'pred'   => 'Non-secretory protein',
 492                 'end'    => '20',
 493                 },
 494        };
 495
 496 # No filters available with hmm on short output
 497 $in      = test_input_file("signalp.hmm.short");
 498 $signalp = Bio::Tools::Signalp::ExtendedSignalp->new(
 499                                                      -file => $in,
 500                                                     );
 501
 502 ok($signalp);
 503 $i = 1;
 504
 505 while(my $feat = $signalp->next_feature()){
 506     is($feat->seq_id(), $res->{$i}->{id});
 507     is($feat->end(), $res->{$i}->{end});
 508     is(($feat->get_tag_values('peptideProb'))[0], $res->{$i}->{prob});
 509     is(($feat->get_tag_values('cleavageSiteProb'))[0], $res->{$i}->{cleav});
 510     is(($feat->get_tag_values('signalpPrediction'))[0], $res->{$i}->{pred});
 511     $i++;
 512 }
 513
 514
 515 exit 0;