lib/CXGN/Transcript/CDS.pm

   1
   2 =head1 NAME
   3
   4 CXGN::Transcript::CDS - a class that deals with coding sequence associated with unigenes.
   5
   6 =head1 DESCRIPTION
   7
   8 Stores predicted CDS and protein sequences from unigene data. The two methods that are currently used are ESTScan and a simple longest 6 frame translation.
   9
  10 The ESTScan CDS predictions contain edits. Insertions are reflected by inserted X's into the sequence, while removed nucleotides are represented by lower case letters in an uppercase background. This sequence is stored in the seq_text field. The actual cds sequence with the lower case letters removed is stored in the seq_edits table.
  11
  12 In general, the correct cds sequence is stored into cds.cds_seq column, and is available through the get_cds_seq accessor.
  13 The corresponding protein sequence is available through get_protein_seq. Note that ESTScan has a bug in the way it calculates the protein (ignores N's in the nucleotide sequence); therefore, protein sequences need to be calculated using the the cds_translate.pl script in /sgn-tools/unigene.
  14
  15 The longest 6-frame translations can be generated using the get_longest_protein.pl script in /sgn-tools/protein. It returns both the cds and the protein sequence in two separate fasta files.
  16
  17
  18 =head1 AUTHOR
  19
  20 Lukas Mueller <lam87@cornell.edu>
  21
  22 =head1 EDITS
  23
  24 reformat_unigene_sequence_with_edit() added 06/23/08 by Mallory Freeberg
  25
  26 =head1 METHODS
  27
  28 This class defines the following methods:
  29
  30 =cut
  31
  32 use strict;
  33
  34 package CXGN::Transcript::CDS;
  35
  36 use base qw | CXGN::DB::Object |;
  37 use base qw | Bio::Seq |;
  38 use base qw | CXGN::Transcript::Unigene |;
  39
  40 =head2 new
  41
  42  Usage:
  43  Desc:
  44  Ret:
  45  Args:
  46  Side Effects:
  47  Example:
  48
  49 =cut
  50
  51 sub new {
  52     my $class = shift;
  53     my $dbh = shift;
  54     my $id = shift;
  55     my $self = $class->SUPER::new($dbh);
  56     if ($id) {
  57         $self->set_cds_id($id);
  58         $self->fetch();
  59     }
  60     return $self;
  61 }
  62
  63 =head2 new_with_unigene_id
  64
  65  Usage:
  66  Desc:
  67  Ret:
  68  Args:
  69  Side Effects:
  70  Example:
  71
  72 =cut
  73
  74 sub new_with_unigene_id {
  75     my $class = shift;
  76     my $dbh = shift;
  77     my $unigene_id = shift;
  78     my $query = "SELECT cds_id FROM cds WHERE unigene_id = ?";
  79     my $sth = $dbh->prepare($query);
  80     $sth->execute($unigene_id);
  81     my ($cds_id) = $sth->fetchrow_array();
  82
  83     my $self = CXGN::Transcript::CDS->new($dbh, $cds_id);
  84     return $self;
  85
  86 }
  87
  88 =head2 exists
  89
  90  Usage:
  91  Desc:
  92  Ret:
  93  Args:
  94  Side Effects:
  95  Example:
  96
  97 =cut
  98
  99 sub exists {
 100     my $dbh = shift;
 101     my $unigene_id = shift;
 102     my $type = shift;
 103
 104     my $query = "SELECT cds_id FROM sgn.cds WHERE unigene_id=? and type=?";
 105
 106     my $sth = $dbh->prepare($query);
 107     $sth->execute($unigene_id, $type);
 108
 109     my @cds_ids = ();
 110     while (my ($cds_id) = $sth->fetchrow_array()) {
 111         push @cds_ids, $cds_id;
 112     }
 113     return @cds_ids;
 114
 115 }
 116
 117
 118
 119 =head2 fetch
 120
 121  Usage:
 122  Desc:
 123  Ret:
 124  Args:
 125  Side Effects:
 126  Example:
 127
 128 =cut
 129
 130 sub fetch {
 131     my $self = shift;
 132     my $query = "SELECT cds_id, unigene_id, protein_feature_id, seq_text, seq_edits, cds_seq, protein_seq, cds.\"begin\", cds.\"end\", cds.forward_reverse, frame, cds.run_id, cds.score, method, preferred FROM sgn.cds WHERE cds_id=?";
 133     my $sth = $self->get_dbh()->prepare($query);
 134     $sth->execute($self->get_cds_id());
 135     my ($cds_id, $unigene_id, $protein_feature_id, $seq_text, $seq_edits, $cds_seq, $protein_seq, $begin, $end, $forward_reverse, $frame, $run_id, $score, $method, $preferred) =
 136         $sth->fetchrow_array();
 137
 138     $self->set_cds_id($cds_id);
 139     $self->set_unigene_id($unigene_id);
 140     $self->set_protein_feature_id($protein_feature_id);
 141     $self->set_cds_seq($cds_seq);
 142     $self->set_seq_text($seq_text);
 143     $self->set_seq_edits($seq_edits);
 144     $self->set_protein_seq($protein_seq);
 145     $self->set_begin($begin);
 146     $self->set_end($end);
 147     $self->set_direction($forward_reverse);
 148     $self->set_frame($frame);
 149     $self->set_run_id($run_id);
 150     $self->set_score($score);
 151     $self->set_method($method);
 152     $self->set_preferred($preferred);
 153
 154     return $cds_id;
 155 }
 156
 157 sub store {
 158     my $self = shift;
 159     if ($self->get_cds_id()) {
 160         my $query = "UPDATE sgn.cds set
 161                      unigene_id=?, protein_feature_id=?, seq_text=?, seq_edits=?, cds_seq=?, protein_seq=?, \"begin\"=?, \"end\"=?, forward_reverse=?, frame=?, run_id=?, score=?, method=?, preferred=?
 162                      WHERE cds_id=?";
 163         my $sth = $self->get_dbh()->prepare($query);
 164         $sth->execute(
 165                        $self->get_unigene_id(),
 166                        $self->get_protein_feature_id(),
 167                        $self->get_seq_text(),
 168                        $self->get_seq_edits(),
 169                        $self->get_cds_seq(),
 170                        $self->get_protein_seq(),
 171                        $self->get_begin(),
 172                        $self->get_end(),
 173                        $self->get_direction(),
 174                        $self->get_frame(),
 175                        $self->get_run_id(),
 176                        $self->get_score(),
 177                        $self->get_method(),
 178                        $self->get_preferred(),
 179                        $self->get_cds_id()
 180
 181
 182                        );
 183
 184     }
 185     else {
 186         my $query = "INSERT INTO sgn.cds (
 187                        unigene_id, protein_feature_id, seq_text, seq_edits, cds_seq, protein_seq, \"begin\", \"end\", forward_reverse, frame, run_id, score, method, preferred) VALUES (?, ?,?, ?, ?, ?, ?, ?, ?, ?,?, ?,?,? )";
 188         my $sth = $self->get_dbh()->prepare($query);
 189         $sth->execute(
 190                         $self->get_unigene_id(),
 191                       $self->get_protein_feature_id(),
 192                        $self->get_seq_text(),
 193                        $self->get_seq_edits(),
 194                       $self->get_cds_seq(),
 195                        $self->get_protein_seq(),
 196                        $self->get_begin(),
 197                        $self->get_end(),
 198                        $self->get_direction(),
 199                       $self->get_frame(),
 200                        $self->get_run_id(),
 201                        $self->get_score(),
 202                       $self->get_method(),
 203                       $self->get_preferred()
 204                       );
 205         return $self->get_currval("sgn.cds_cds_id_seq");
 206     }
 207
 208 }
 209
 210 =head2 accessors get_cds_id, set_cds_id
 211
 212  Usage:
 213  Property:      the primary key of the table.
 214
 215 =cut
 216
 217 sub get_cds_id {
 218   my $self=shift;
 219   return $self->{cds_id};
 220
 221 }
 222
 223 sub set_cds_id {
 224   my $self=shift;
 225   $self->{cds_id}=shift;
 226 }
 227
 228 =head2 accessors get_unigene_id, set_unigene_id
 229
 230  Usage:
 231  Property:     foreign key to unigene. The unigene this
 232                cds entry is associated with.
 233  Ret:
 234  Args:
 235  Side Effects:
 236  Example:
 237
 238 =cut
 239
 240 sub get_unigene_id {
 241   my $self=shift;
 242   return $self->{unigene_id};
 243
 244 }
 245
 246 sub set_unigene_id {
 247   my $self=shift;
 248   $self->{unigene_id}=shift;
 249 }
 250
 251 =head2 accessors get_protein_feature_id, set_protein_feature_id
 252
 253  Usage:
 254  Desc:
 255  Property
 256  Side Effects:
 257  Example:
 258
 259 =cut
 260
 261 sub get_protein_feature_id {
 262   my $self = shift;
 263   return $self->{protein_feature_id};
 264 }
 265
 266 sub set_protein_feature_id {
 267   my $self = shift;
 268   $self->{protein_feature_id} = shift;
 269 }
 270
 271
 272 =head2 accessors get_seq_text, set_seq_text
 273
 274  Usage:        my $seq_text = $cds->get_seq_text()
 275  Desc:         gets the sequence text predicted by ESTScan.
 276                BIG CAVEAT: this sequence contains the nucleotides
 277                that ESTScan removed in lower case, while everything
 278                else is in uppercase. Use get_cds_seq() to get the
 279                correct cds sequence irrespective of method used.
 280  Side Effects:
 281  Example:
 282
 283 =cut
 284
 285 sub get_seq_text {
 286   my $self=shift;
 287   return $self->{seq_text};
 288
 289 }
 290
 291 sub set_seq_text {
 292   my $self=shift;
 293   $self->{seq_text}=shift;
 294 }
 295
 296 =head2 accessors get_cds_seq, set_cds_seq
 297
 298  Usage:        my $cds_seq = $cds->get_cds_seq
 299  Desc:         gets the cds sequence
 300  Side Effects: property usually set in constructor...
 301                the value is populated from the cds_seq field
 302                of the sgn.cds database table. This field contains
 303                the correct cds sequence irregardless of the method
 304                used.
 305  Example:
 306
 307 =cut
 308
 309 sub get_cds_seq {
 310   my $self=shift;
 311   return $self->{cds_seq};
 312
 313 }
 314
 315
 316 sub set_cds_seq {
 317   my $self=shift;
 318   $self->{cds_seq}=shift;
 319 }
 320
 321
 322
 323 =head2 accessors get_seq_edits, set_seq_edits
 324
 325  Usage:        my $seq_edits = $cds->get_seq_edits
 326  Desc:         gets the sequence that ESTScan predicted
 327  Ret:          with the lower case letter removed.
 328                To access the cds irregardless of method used,
 329                use the get_cds_seq() accessor.
 330  Args:
 331  Side Effects:
 332  Example:
 333
 334 =cut
 335
 336 sub get_seq_edits {
 337   my $self=shift;
 338   return $self->{seq_edits};
 339
 340 }
 341
 342 sub set_seq_edits {
 343   my $self=shift;
 344   $self->{seq_edits}=shift;
 345 }
 346
 347 =head2 accessors get_protein_seq, set_protein_seq
 348
 349  Usage:
 350  Desc:
 351  Ret:
 352  Args:
 353  Side Effects:
 354  Example:
 355
 356 =cut
 357
 358 sub get_protein_seq {
 359     my $self=shift;
 360     return $self->{protein_seq};
 361
 362 }
 363
 364 sub set_protein_seq {
 365   my $self=shift;
 366   $self->{protein_seq}=shift;
 367 }
 368
 369 =head2 accessors get_begin, set_begin
 370
 371  Usage:
 372  Desc:
 373  Ret:
 374  Args:
 375  Side Effects:
 376  Example:
 377
 378 =cut
 379
 380 sub get_begin {
 381   my $self=shift;
 382   return $self->{begin};
 383
 384 }
 385
 386 sub set_begin {
 387   my $self=shift;
 388   $self->{begin}=shift;
 389 }
 390
 391 =head2 accessors get_end, set_end
 392
 393  Usage:
 394  Desc:
 395  Ret:
 396  Args:
 397  Side Effects:
 398  Example:
 399
 400 =cut
 401
 402 sub get_end {
 403   my $self=shift;
 404   return $self->{end};
 405
 406 }
 407
 408 sub set_end {
 409   my $self=shift;
 410   $self->{end}=shift;
 411 }
 412
 413 # =head2 accessors get_forward_reverse, set_forward_reverse
 414
 415 #  Usage:
 416 #  Desc:
 417 #  Ret:
 418 #  Args:
 419 #  Side Effects:
 420 #  Example:
 421
 422 # =cut
 423
 424 # sub get_forward_reverse {
 425 #   my $self=shift;
 426 #   return $self->{forward_reverse};
 427
 428 # }
 429
 430 # sub set_forward_reverse {
 431 #   my $self=shift;
 432 #   $self->{forward_reverse}=shift;
 433 # }
 434
 435
 436 =head2 get_direction
 437
 438  Usage:        my $direction = $cds->get_direction()
 439  Desc:         gets the direction of the cds relative to the
 440                unigene. Either "F" for forward or "R" for
 441                reverse.
 442  Side Effects: this property is set in the constructor.
 443                it maps to the somewhat cumbersomly named
 444                forward_reverse column in the database.
 445  Example:
 446
 447 =cut
 448
 449 sub get_direction {
 450   my $self=shift;
 451   return $self->{direction};
 452
 453 }
 454
 455 sub set_direction {
 456   my $self=shift;
 457   $self->{direction}=shift;
 458 }
 459
 460 =head2 accessors get_frame, set_frame
 461
 462  Usage:
 463  Desc:
 464  Ret:
 465  Args:
 466  Side Effects:
 467  Example:
 468
 469 =cut
 470
 471 sub get_frame {
 472   my $self=shift;
 473   return $self->{frame};
 474
 475 }
 476
 477
 478 sub set_frame {
 479     my $self=shift;
 480     $self->{frame}=shift;
 481 }
 482
 483
 484
 485 =head2 accessors get_run_id, set_run_id
 486
 487  Usage:
 488  Desc:
 489  Ret:
 490  Args:
 491  Side Effects:
 492  Example:
 493
 494 =cut
 495
 496 sub get_run_id {
 497   my $self=shift;
 498   return $self->{run_id};
 499
 500 }
 501
 502 sub set_run_id {
 503   my $self=shift;
 504   $self->{run_id}=shift;
 505 }
 506
 507 =head2 accessors get_score, set_score
 508
 509  Usage:
 510  Desc:
 511  Ret:
 512  Args:
 513  Side Effects:
 514  Example:
 515
 516 =cut
 517
 518 sub get_score {
 519   my $self=shift;
 520   return $self->{score};
 521
 522 }
 523
 524 sub set_score {
 525   my $self=shift;
 526   $self->{score}=shift;
 527 }
 528
 529 # =head2 accessors get_longest_frame_translation, set_longest_frame_translation
 530
 531 #  Usage:
 532 #  Desc:
 533 #  Ret:
 534 #  Args:
 535 #  Side Effects:
 536 #  Example:
 537
 538 # =cut
 539
 540 # sub get_longest_frame_translation {
 541 #   my $self=shift;
 542 #   return $self->{longest_frame_translation};
 543
 544 # }
 545
 546 # sub set_longest_frame_translation {
 547 #   my $self=shift;
 548 #   $self->{longest_frame_translation}=shift;
 549 # }
 550
 551 # =head2 accessors get_longest_frame, set_longest_frame
 552
 553 #  Usage:
 554 #  Desc:
 555 #  Ret:
 556 #  Args:
 557 #  Side Effects:
 558 #  Example:
 559
 560 # =cut
 561
 562 # sub get_longest_frame {
 563 #   my $self=shift;
 564 #   return $self->{longest_frame};
 565
 566 # }
 567
 568 # sub set_longest_frame {
 569 #   my $self=shift;
 570 #   $self->{longest_frame}=shift;
 571 # }
 572
 573 # =head2 function get_proteins
 574
 575 #  Usage:
 576 #  Desc:
 577 #  Ret:
 578 #  Args:
 579 #  Side Effects:
 580 #  Example:
 581
 582 # =cut
 583
 584 # sub get_proteins {
 585 #     my $self = shift;
 586 #     my $sgn = $self->get_dbh()->qualify_schema("sgn");
 587 #     my $query = "SELECT protein_id FROM $sgn.protein WHERE cds_id=?";
 588 #     my $sth = $self->get_dbh()->prepare($query);
 589 #     $sth->execute($self->get_cds_id());
 590 #     my @proteins = ();
 591 #     while (my ($protein_id) = $sth->fetchrow_array()) {
 592 #       push @proteins, CXGN::Transcript::Protein->new($self->get_dbh(), $protein_id);
 593 #     }
 594 #     return @proteins;
 595 # }
 596
 597
 598 =head2 accessors get_method, set_method
 599
 600  Usage:        my $method = $cds->get_method()
 601  Desc:         the method used to predict the cds/protein seq
 602  Ret:          either "estscan" or "longest6frame"
 603  Args:         none
 604
 605 =cut
 606
 607 sub get_method {
 608     my $self=shift;
 609     return $self->{method};
 610
 611 }
 612
 613 sub set_method {
 614     my $self=shift;
 615     $self->{method}=shift;
 616 }
 617
 618
 619 =head2 get_preferred
 620
 621  Usage:
 622  Desc:
 623  Ret:
 624  Args:
 625  Side Effects:
 626  Example:
 627
 628 =cut
 629
 630 sub get_preferred {
 631     my $self=shift;
 632
 633     return $self->{preferred};
 634
 635 }
 636
 637 =head2 set_preferred
 638
 639  Usage:
 640  Desc:
 641  Ret:
 642  Args:
 643  Side Effects:
 644  Example:
 645
 646 =cut
 647
 648 sub set_preferred {
 649     my $self=shift;
 650     my $preferred = shift;
 651     if ($preferred=~/t/i) {
 652         $preferred = 1;
 653     }
 654     if ($preferred=~/f/i) {
 655         $preferred= 0;
 656     }
 657     $self->{preferred}=$preferred;
 658 }
 659
 660 =head2 get_signalp_info
 661
 662  Usage:        my ($nn_ypos, $nn_score, $nn_d) = $unigene->get_signalp_info()
 663  Desc:
 664  Ret:
 665  Args:
 666  Side Effects:
 667  Example:
 668
 669 =cut
 670
 671 sub get_signalp_info {
 672     my $self = shift;
 673     my $query = "SELECT nn_ypos, nn_score, nn_d
 674         FROM public.signalp
 675         WHERE cds_id=? ";
 676     my $sth = $self->get_dbh()->prepare($query);
 677     $sth->execute($self->get_cds_id());
 678     my ($nn_ypos, $nn_score, $nn_d) = $sth->fetchrow_array();
 679     return ($nn_d, $nn_ypos, $nn_score);
 680 }
 681
 682
 683 =head2 function get_interpro_domains
 684
 685  Usage:
 686  Desc:
 687  Ret:
 688  Args:
 689  Side Effects:
 690  Example:
 691
 692 =cut
 693
 694 sub get_interpro_domains {
 695     my $self = shift;
 696     my $query = "SELECT interpro_accession, interpro.description, match_begin, match_end
 697                                                 FROM sgn.interpro
 698                                                 LEFT JOIN sgn.domain USING (interpro_id)
 699                                                 LEFT JOIN sgn.domain_match USING (domain_id)
 700                                                 WHERE cds_id = ?
 701                                                         AND hit_status LIKE 'T' ";
 702     my $sth = $self->get_dbh()->prepare_cached($query);
 703     $sth->execute($self->get_cds_id());
 704     my @interpro_domain_list = ();
 705     while (my ($interpro_accession, $description, $match_begin, $match_end)= $sth->fetchrow_array()) {
 706         push @interpro_domain_list, [$interpro_accession,$description, $match_begin, $match_end];
 707     }
 708     return @interpro_domain_list;
 709
 710 }
 711
 712 return 1;