t/Signalp2.t

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