Bio/Variation/VariantI.pm

   1 #
   2 # BioPerl module for Bio::Variation::VariantI
   3 #
   4 # Please direct questions and support issues to <bioperl-l@bioperl.org>
   5 #
   6 # Cared for by Heikki Lehvaslaiho <heikki-at-bioperl-dot-org>
   7 #
   8 # Copyright Heikki Lehvaslaiho
   9 #
  10 # You may distribute this module under the same terms as perl itself
  11
  12 # POD documentation - main docs before the code
  13
  14 =head1 NAME
  15
  16 Bio::Variation::VariantI - Sequence Change SeqFeature abstract class
  17
  18 =head1 SYNOPSIS
  19
  20   #get Bio::Variant::VariantI somehow
  21   print $var->restriction_changes, "\n";
  22   foreach $allele ($var->each_Allele) {
  23       #work on Bio::Variation::Allele objects
  24   }
  25
  26 =head1 DESCRIPTION
  27
  28 This superclass defines common methods to basic sequence changes.  The
  29 instantiable classes Bio::Variation::DNAMutation,
  30 Bio::Variation::RNAChange and Bio::Variation::AAChange use them.
  31 See L<Bio::Variation::DNAMutation>, L<Bio::Variation::RNAChange>,
  32 and L<Bio::Variation::AAChange> for more information.
  33
  34 These classes store information, heavy computation to detemine allele
  35 sequences is done elsewhere.
  36
  37 The database cross-references are implemented as
  38 Bio::Annotation::DBLink objects. The methods to access them are
  39 defined in Bio::DBLinkContainerI. See L<Bio::Annotation::DBLink>
  40 and L<Bio::DBLinkContainerI> for details.
  41
  42 Bio::Variation::VariantI redifines and extends
  43 Bio::SeqFeature::Generic for sequence variations. This class
  44 describes specific sequence change events. These events are always
  45 from a specific reference sequence to something different. See
  46 L<Bio::SeqFeature::Generic> for more information.
  47
  48 IMPORTANT: The notion of reference sequence permeates all
  49 Bio::Variation classes. This is especially important to remember when
  50 dealing with Alleles. In a polymorphic site, there can be a large
  51 number of alleles. One of then has to be selected to be the reference
  52 allele (allele_ori). ALL the rest has to be passed to the Variant
  53 using the method add_Allele, including the mutated allele in a
  54 canonical mutation. The IO modules and generated attributes depend on
  55 it. They ignore the allele linked to using allele_mut and circulate
  56 each Allele returned by each_Allele into allele_mut and calculate
  57 the changes between that and allele_ori.
  58
  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 the
  66 Bioperl mailing lists  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 the bugs and their resolution.  Bug reports can be submitted via the
  86 web:
  87
  88   https://github.com/bioperl/bioperl-live/issues
  89
  90 =head1 AUTHOR - Heikki Lehvaslaiho
  91
  92 Email:  heikki-at-bioperl-dot-org
  93
  94 =head1 APPENDIX
  95
  96 The rest of the documentation details each of the object
  97 methods. Internal methods are usually preceded with a _
  98
  99 =cut
 100
 101
 102 # Let the code begin...
 103
 104
 105 package Bio::Variation::VariantI;
 106 use strict;
 107 # Object preamble - inheritance
 108
 109 use base qw(Bio::Root::Root Bio::SeqFeature::Generic Bio::DBLinkContainerI);
 110
 111 =head2 id
 112
 113  Title   : id
 114  Usage   : $obj->id
 115  Function:
 116
 117            Read only method. Returns the id of the variation object.
 118            The id is the id of the first DBLink object attached to this object.
 119
 120  Example :
 121  Returns : scalar
 122  Args    : none
 123
 124 =cut
 125
 126 sub id {
 127    my ($self) = @_;
 128    my @ids = $self->each_DBLink;
 129    my $id = $ids[0] if scalar @ids > 0;
 130    return $id->database. "::". $id->primary_id if $id;
 131 }
 132
 133
 134 =head2 add_Allele
 135
 136  Title   : add_Allele
 137  Usage   : $self->add_Allele($allele)
 138  Function:
 139
 140             Adds one Bio::Variation::Allele into the list of alleles.
 141             Note that the method forces the convention that nucleotide
 142             sequence is in lower case and amino acds are in upper
 143             case.
 144
 145  Example :
 146  Returns : 1 when succeeds, 0 for failure.
 147  Args    : Allele object
 148
 149 =cut
 150
 151
 152 sub add_Allele {
 153   my ($self,$value) = @_;
 154   if (defined $value) {
 155       if( ! $value->isa('Bio::Variation::Allele') ) {
 156           my $com = ref $value;
 157           $self->throw("Is not a Allele object but a  [$com]");
 158           return 0;
 159       } else {
 160           if ( $self->isa('Bio::Variation::AAChange') ) {
 161               $value->seq( uc $value->seq) if $value->seq;
 162           } else {
 163               $value->seq( lc $value->seq) if $value->seq;
 164           }
 165           push(@{$self->{'alleles'}},$value);
 166           $self->allele_mut($value); #????
 167           return 1;
 168       }
 169   } else {
 170       return 0;
 171   }
 172 }
 173
 174
 175 =head2 each_Allele
 176
 177  Title   : alleles
 178  Usage   : $obj->each_Allele();
 179  Function:
 180
 181              Returns a list of Bio::Variation::Allele objects
 182
 183  Example :
 184  Returns : list of Alleles
 185  Args    : none
 186
 187 =cut
 188
 189 sub each_Allele{
 190    my ($self,@args) = @_;
 191    return @{$self->{'alleles'}};
 192 }
 193
 194
 195 =head2 isMutation
 196
 197  Title   : isMutation
 198  Usage   : print join('/', $obj->each_Allele) if not $obj->isMutation;
 199  Function:
 200
 201            Returns or sets the boolean value indicating that the
 202            variant descibed is a canonical mutation with two alleles
 203            assinged to be the original (wild type) allele and mutated
 204            allele, respectively. If this value is not set, it is
 205            assumed that the Variant descibes polymorphisms.
 206
 207  Returns : a boolean
 208
 209 =cut
 210
 211 sub isMutation {
 212     my ($self,$value) = @_;
 213     if (defined $value) {
 214         if ($value ) {
 215             $self->{'isMutation'} = 1;
 216         } else {
 217             $self->{'isMutation'} = 0;
 218         }
 219     }
 220     return $self->{'isMutation'};
 221 }
 222
 223
 224 =head2 allele_ori
 225
 226  Title   : allele_ori
 227  Usage   : $obj->allele_ori();
 228  Function:
 229
 230             Links to and returns the Bio::Variation::Allele object.
 231             If value is not set, returns false. All other Alleles are
 232             compared to this.
 233
 234             Amino acid sequences are stored in upper case characters,
 235             others in lower case.
 236
 237  Example :
 238  Returns : string
 239  Args    : string
 240
 241 See L<Bio::Variation::Allele> for more.
 242
 243 =cut
 244
 245 sub allele_ori {
 246    my ($self,$value) = @_;
 247    if( defined $value) {
 248        if ( ! ref $value || ! $value->isa('Bio::Variation::Allele')) {
 249            $self->throw("Value is not Bio::Variation::Allele but [$value]");
 250        } else {
 251            if ( $self->isa('Bio::Variation::AAChange') ) {
 252                $value->seq( uc $value->seq) if $value->seq;
 253            } else {
 254                $value->seq( lc $value->seq) if $value->seq;
 255            }
 256            $self->{'allele_ori'} = $value;
 257        }
 258    }
 259    return $self->{'allele_ori'};
 260 }
 261
 262
 263 =head2 allele_mut
 264
 265  Title   : allele_mut
 266  Usage   : $obj->allele_mut();
 267  Function:
 268
 269              Links to and returns the Bio::Variation::Allele
 270              object.  Sets and returns the mutated allele sequence.
 271              If value is not set, returns false.
 272
 273              Amino acid sequences are stored in upper case characters,
 274              others in lower case.
 275
 276  Example :
 277  Returns : string
 278  Args    : string
 279
 280 See L<Bio::Variation::Allele> for more.
 281
 282 =cut
 283
 284
 285 sub allele_mut {
 286    my ($self,$value) = @_;
 287    if( defined $value) {
 288        if ( ! ref $value || ! $value->isa('Bio::Variation::Allele')) {
 289            $self->throw("Value is not Bio::Variation::Allele but [$value]");
 290        } else {
 291            if ( $self->isa('Bio::Variation::AAChange') ) {
 292                $value->seq( uc $value->seq) if $value->seq;
 293            } else {
 294                $value->seq( lc $value->seq) if $value->seq;
 295            }
 296            $self->{'allele_mut'} = $value;
 297        }
 298    }
 299    return $self->{'allele_mut'};
 300 }
 301
 302 =head2 length
 303
 304  Title   : length
 305  Usage   : $obj->length();
 306  Function:
 307
 308             Sets and returns the length of the affected original
 309             allele sequence.  If value is not set, returns false == 0.
 310
 311             Value 0 means that the variant position is before the
 312             start=end sequence position. (Value 1 would denote a point
 313             mutation). This follows the convension to report an
 314             insertion (2insT) in equivalent way to a corresponding
 315             deletion (2delT) (Think about indel polymorpism ATC <=> AC
 316             where the origianal state is not known ).
 317
 318  Example :
 319  Returns : string
 320  Args    : string
 321
 322 =cut
 323
 324
 325 sub length {
 326    my ($self,$value) = @_;
 327    if ( defined $value) {
 328        $self->{'length'} = $value;
 329   }
 330    if ( ! exists $self->{'length'} ) {
 331        return 0;
 332    }
 333    return $self->{'length'};
 334 }
 335
 336 =head2 upStreamSeq
 337
 338  Title   : upStreamSeq
 339  Usage   : $obj->upStreamSeq();
 340  Function:
 341
 342             Sets and returns upstream flanking sequence string.  If
 343             value is not set, returns false. The sequence should be
 344             >=25 characters long, if possible.
 345
 346  Example :
 347  Returns : string or false
 348  Args    : string
 349
 350 =cut
 351
 352
 353 sub upStreamSeq {
 354     my ($self,$value) = @_;
 355     if( defined $value) {
 356         $self->{'upstreamseq'} = $value;
 357     }
 358    return $self->{'upstreamseq'};
 359 }
 360
 361
 362 =head2 dnStreamSeq
 363
 364  Title   : dnStreamSeq
 365  Usage   : $obj->dnStreamSeq();
 366  Function:
 367
 368             Sets and returns dnstream flanking sequence string.  If
 369             value is not set, returns false. The sequence should be
 370             >=25 characters long, if possible.
 371
 372  Example :
 373  Returns : string or false
 374  Args    : string
 375
 376 =cut
 377
 378
 379 sub dnStreamSeq {
 380     my ($self,$value) = @_;
 381     if( defined $value) {
 382         $self->{'dnstreamseq'} = $value;
 383     }
 384     return $self->{'dnstreamseq'};
 385
 386 }
 387
 388
 389 =head2 label
 390
 391  Title   : label
 392  Usage   : $obj->label();
 393  Function:
 394
 395             Sets and returns mutation event label(s).  If value is not
 396             set, or no argument is given returns false.  Each
 397             instantiable class needs to implement this method. Valid
 398             values are listed in 'Mutation event controlled vocabulary' in
 399             http://www.ebi.ac.uk/mutations/recommendations/mutevent.html.
 400
 401  Example :
 402  Returns : string
 403  Args    : string
 404
 405 =cut
 406
 407
 408 sub label {
 409     my ($self,$value) = @_;
 410     $self->throw_not_implemented();
 411 }
 412
 413
 414
 415 =head2 status
 416
 417  Title   : status
 418  Usage   : $obj->status()
 419  Function:
 420
 421            Returns the status of the sequence change object.
 422            Valid values are: 'suspected' and 'proven'
 423
 424  Example : $obj->status('proven');
 425  Returns : scalar
 426  Args    : valid string (optional, for setting)
 427
 428
 429 =cut
 430
 431
 432 sub status {
 433    my ($self,$value) = @_;
 434    my %status = (suspected => 1,
 435                  proven => 1
 436                  );
 437
 438    if( defined $value) {
 439        $value = lc $value;
 440        if ($status{$value}) {
 441            $self->{'status'} = $value;
 442        }
 443        else {
 444            $self->throw("$value is not valid status value!");
 445        }
 446     }
 447    if( ! exists $self->{'status'} ) {
 448        return "$self";
 449    }
 450    return $self->{'status'};
 451 }
 452
 453
 454 =head2 proof
 455
 456  Title   : proof
 457  Usage   : $obj->proof()
 458  Function:
 459
 460            Returns the proof of the sequence change object.
 461            Valid values are: 'computed' and 'experimental'.
 462
 463  Example : $obj->proof('computed');
 464  Returns : scalar
 465  Args    : valid string (optional, for setting)
 466
 467
 468 =cut
 469
 470
 471 sub proof {
 472     my ($self,$value) = @_;
 473     my %proof = (computed => 1,
 474                  experimental => 1
 475                  );
 476
 477     if( defined $value) {
 478         $value = lc $value;
 479         if ($proof{$value}) {
 480             $self->{'proof'} = $value;
 481         } else {
 482             $self->throw("$value is not valid proof value!");
 483         }
 484     }
 485     return $self->{'proof'};
 486 }
 487
 488
 489 =head2 region
 490
 491  Title   : region
 492  Usage   : $obj->region();
 493  Function:
 494
 495             Sets and returns the name of the sequence region type or
 496             protein domain at this location.  If value is not set,
 497             returns false.
 498
 499  Example :
 500  Returns : string
 501  Args    : string
 502
 503 =cut
 504
 505
 506 sub region {
 507     my ($self,$value) = @_;
 508     if( defined $value) {
 509         $self->{'region'} = $value;
 510     }
 511     return $self->{'region'};
 512 }
 513
 514
 515 =head2 region_value
 516
 517  Title   : region_value
 518  Usage   : $obj->region_value();
 519  Function:
 520
 521             Sets and returns the name of the sequence region_value or
 522             protein domain at this location.  If value is not set,
 523             returns false.
 524
 525  Example :
 526  Returns : string
 527  Args    : string
 528
 529 =cut
 530
 531
 532 sub region_value {
 533     my ($self,$value) = @_;
 534     if( defined $value) {
 535         $self->{'region_value'} = $value;
 536     }
 537     return $self->{'region_value'};
 538 }
 539
 540 =head2 region_dist
 541
 542  Title   : region_dist
 543  Usage   : $obj->region_dist();
 544  Function:
 545
 546             Sets and returns the distance tot the closest region
 547             (i.e. intro/exon or domain) boundary. If distance is not
 548             set, returns false.
 549
 550  Example :
 551  Returns : integer
 552  Args    : integer
 553
 554 =cut
 555
 556
 557 sub region_dist {
 558     my ($self,$value) = @_;
 559     if( defined $value) {
 560        if (  not $value =~ /^[+-]?\d+$/ ) {
 561            $self->throw("[$value] for region_dist has to be an integer\n");
 562         } else {
 563             $self->{'region_dist'} = $value;
 564         }
 565     }
 566     return $self->{'region_dist'};
 567 }
 568
 569
 570 =head2 numbering
 571
 572  Title   : numbering
 573  Usage   : $obj->numbering()
 574  Function:
 575
 576            Returns the numbering chema used locating sequnce features.
 577            Valid values are: 'entry' and 'coding'
 578
 579  Example : $obj->numbering('coding');
 580  Returns : scalar
 581  Args    : valid string (optional, for setting)
 582
 583
 584 =cut
 585
 586
 587 sub numbering {
 588    my ($self,$value) = @_;
 589    my %numbering = (entry => 1,
 590                     coding => 1
 591                     );
 592
 593    if( defined $value) {
 594        $value = lc $value;
 595        if ($numbering{$value}) {
 596            $self->{'numbering'} = $value;
 597        }
 598        else {
 599            $self->throw("'$value' is not a valid for numbering!");
 600        }
 601     }
 602    if( ! exists $self->{'numbering'} ) {
 603        return "$self";
 604    }
 605    return $self->{'numbering'};
 606 }
 607
 608 =head2 mut_number
 609
 610  Title   : mut_number
 611  Usage   : $num = $obj->mut_number;
 612          : $num = $obj->mut_number($number);
 613  Function:
 614
 615            Returns or sets the number identifying the order in which the
 616            mutation has been issued. Numbers shouldstart from 1.
 617            If the number has never been set, the method will return ''
 618
 619            If you want the output from IO modules look nice and, for
 620            multivariant/allele variations, make sense you better set
 621            this attribute.
 622
 623  Returns : an integer
 624
 625 =cut
 626
 627
 628 sub mut_number {
 629     my ($self,$value) = @_;
 630     if (defined $value) {
 631         $self->{'mut_number'} = $value;
 632     }
 633     unless (exists $self->{'mut_number'}) {
 634         return ('');
 635     } else {
 636         return $self->{'mut_number'};
 637     }
 638 }
 639
 640
 641 =head2 SeqDiff
 642
 643  Title   : SeqDiff
 644  Usage   : $mutobj = $obj->SeqDiff;
 645          : $mutobj = $obj->SeqDiff($objref);
 646  Function:
 647
 648            Returns or sets the link-reference to the umbrella
 649            Bio::Variation::SeqDiff object.  If there is no link,
 650            it will return undef
 651
 652            Note: Adding a variant into a SeqDiff object will
 653            automatically set this value.
 654
 655  Returns : an obj_ref or undef
 656
 657 See L<Bio::Variation::SeqDiff> for more information.
 658
 659 =cut
 660
 661 sub SeqDiff {
 662     my ($self,$value) = @_;
 663     if (defined $value) {
 664         if( ! $value->isa('Bio::Variation::SeqDiff') ) {
 665             $self->throw("Is not a Bio::Variation::SeqDiff object but a [$value]");
 666             return;
 667         }
 668         else {
 669             $self->{'seqDiff'} = $value;
 670         }
 671     }
 672     unless (exists $self->{'seqDiff'}) {
 673         return;
 674     } else {
 675         return $self->{'seqDiff'};
 676     }
 677 }
 678
 679 =head2 add_DBLink
 680
 681  Title   : add_DBLink
 682  Usage   : $self->add_DBLink($ref)
 683  Function: adds a link object
 684  Example :
 685  Returns :
 686  Args    :
 687
 688
 689 =cut
 690
 691
 692 sub add_DBLink{
 693    my ($self,$com) = @_;
 694    if( $com && ! $com->isa('Bio::Annotation::DBLink') ) {
 695        $self->throw("Is not a link object but a  [$com]");
 696    }
 697    $com && push(@{$self->{'link'}},$com);
 698 }
 699
 700 =head2 each_DBLink
 701
 702  Title   : each_DBLink
 703  Usage   : foreach $ref ( $self->each_DBlink() )
 704  Function: gets an array of DBlink of objects
 705  Example :
 706  Returns :
 707  Args    :
 708
 709
 710 =cut
 711
 712 sub each_DBLink{
 713    my ($self) = @_;
 714
 715    return @{$self->{'link'}};
 716 }
 717
 718 =head2 restriction_changes
 719
 720  Title   : restriction_changes
 721  Usage   : $obj->restriction_changes();
 722  Function:
 723
 724             Returns a string containing a list of restriction
 725             enzyme changes of form +EcoRI, separated by
 726             commas. Strings need to be valid restriction enzyme names
 727             as stored in REBASE. allele_ori and allele_mut need to be assigned.
 728
 729  Example :
 730  Returns : string
 731  Args    : string
 732
 733 =cut
 734
 735 sub restriction_changes {
 736     my ($self) = @_;
 737
 738     if (not $self->{'re_changes'}) {
 739         my %re = &_enzymes;
 740
 741         # complain if used on AA data
 742         if ($self->isa('Bio::Variation::AAChange')) {
 743             $self->throw('Restriction enzymes do not bite polypeptides!');
 744         }
 745
 746         #sanity checks
 747         $self->warn('Upstream sequence is empty!')
 748             if $self->upStreamSeq eq '';
 749         $self->warn('Downstream sequence is empty!')
 750             if $self->dnStreamSeq eq '';
 751 #        $self->warn('Original allele sequence is empty!')
 752 #            if $self->allele_ori eq '';
 753 #        $self->warn('Mutated allele sequence is empty!')
 754 #            if $self->allele_mut eq '';
 755
 756         #reuse the non empty DNA level list at RNA level if the flanks are identical
 757         #Hint: Check DNAMutation object first
 758         if ($self->isa('Bio::Variation::RNAChange') and  $self->DNAMutation and
 759             $self->upStreamSeq eq $self->DNAMutation->upStreamSeq  and
 760             $self->dnStreamSeq eq $self->DNAMutation->dnStreamSeq and
 761             $self->DNAMutation->restriction_changes ne '' ) {
 762             $self->{'re_changes'} = $self->DNAMutation->restriction_changes;
 763         } else {
 764
 765             #maximum length of a type II restriction site in the current REBASE
 766             my ($le_dn) = 15;
 767             my ($le_up) = $le_dn;
 768
 769             #reduce the flank lengths if the desired length is not available
 770             $le_dn = CORE::length ($self->dnStreamSeq) if $le_dn > CORE::length ($self->dnStreamSeq);
 771             $le_up = CORE::length ($self->upStreamSeq) if $le_up > CORE::length ($self->upStreamSeq);
 772
 773             #Build sequence strings to compare
 774             my ($oriseq, $mutseq);
 775             $oriseq  = $mutseq = substr($self->upStreamSeq, -$le_up, $le_up);
 776             $oriseq .= $self->allele_ori->seq if $self->allele_ori->seq;
 777             $mutseq .= $self->allele_mut->seq if $self->allele_mut->seq;
 778             $oriseq .= substr($self->dnStreamSeq, 0, $le_dn);
 779             $mutseq .= substr($self->dnStreamSeq, 0, $le_dn);
 780
 781             # ... and their reverse complements
 782             my $oriseq_rev = _revcompl ($oriseq);
 783             my $mutseq_rev = _revcompl ($mutseq);
 784
 785             # collect results into a string
 786             my $rec = '';
 787             foreach my $enz (sort keys (%re)) {
 788                 my $site = $re{$enz};
 789                 my @ori = ($oriseq=~ /$site/g);
 790                 my @mut = ($mutseq=~ /$site/g);
 791                 my @ori_r = ($oriseq_rev =~ /$site/g);
 792                 my @mut_r = ($mutseq_rev =~ /$site/g);
 793
 794                 $rec .= '+'. $enz. ", "
 795                     if (scalar @ori < scalar @mut) or (scalar @ori_r < scalar @mut_r);
 796                 $rec .= '-'. $enz. ", "
 797                     if (scalar @ori > scalar @mut) or (scalar @ori_r > scalar @mut_r);
 798
 799             }
 800             $rec = substr($rec, 0, CORE::length($rec) - 2) if $rec ne '';
 801             $self->{'re_changes'} =  $rec;
 802         }
 803     }
 804     return $self->{'re_changes'}
 805 }
 806
 807
 808 sub _revcompl {
 809     # side effect: lower case letters
 810     my ($seq) = shift;
 811
 812     $seq = lc $seq;
 813     $seq =~ tr/acgtrymkswhbvdnx/tgcayrkmswdvbhnx/;
 814     return CORE::reverse $seq;
 815 }
 816
 817
 818 sub _enzymes {
 819  #REBASE version 005   type2.005
 820  my %enzymes =  (
 821          'AarI' => 'cacctgc',
 822          'AatII' => 'gacgtc',
 823          'AccI' => 'gt[ac][gt]ac',
 824          'AceIII' => 'cagctc',
 825          'AciI' => 'ccgc',
 826          'AclI' => 'aacgtt',
 827          'AcyI' => 'g[ag]cg[ct]c',
 828          'AflII' => 'cttaag',
 829          'AflIII' => 'ac[ag][ct]gt',
 830          'AgeI' => 'accggt',
 831          'AhaIII' => 'tttaaa',
 832          'AloI' => 'gaac[acgt][acgt][acgt][acgt][acgt][acgt]tcc',
 833          'AluI' => 'agct',
 834          'AlwNI' => 'cag[acgt][acgt][acgt]ctg',
 835          'ApaBI' => 'gca[acgt][acgt][acgt][acgt][acgt]tgc',
 836          'ApaI' => 'gggccc',
 837          'ApaLI' => 'gtgcac',
 838          'ApoI' => '[ag]aatt[ct]',
 839          'AscI' => 'ggcgcgcc',
 840          'AsuI' => 'gg[acgt]cc',
 841          'AsuII' => 'ttcgaa',
 842          'AvaI' => 'c[ct]cg[ag]g',
 843          'AvaII' => 'gg[at]cc',
 844          'AvaIII' => 'atgcat',
 845          'AvrII' => 'cctagg',
 846          'BaeI' => 'ac[acgt][acgt][acgt][acgt]gta[ct]c',
 847          'BalI' => 'tggcca',
 848          'BamHI' => 'ggatcc',
 849          'BbvCI' => 'cctcagc',
 850          'BbvI' => 'gcagc',
 851          'BbvII' => 'gaagac',
 852          'BccI' => 'ccatc',
 853          'Bce83I' => 'cttgag',
 854          'BcefI' => 'acggc',
 855          'BcgI' => 'cga[acgt][acgt][acgt][acgt][acgt][acgt]tgc',
 856          'BciVI' => 'gtatcc',
 857          'BclI' => 'tgatca',
 858          'BetI' => '[at]ccgg[at]',
 859          'BfiI' => 'actggg',
 860          'BglI' => 'gcc[acgt][acgt][acgt][acgt][acgt]ggc',
 861          'BglII' => 'agatct',
 862          'BinI' => 'ggatc',
 863          'BmgI' => 'g[gt]gccc',
 864          'BplI' => 'gag[acgt][acgt][acgt][acgt][acgt]ctc',
 865          'Bpu10I' => 'cct[acgt]agc',
 866          'BsaAI' => '[ct]acgt[ag]',
 867          'BsaBI' => 'gat[acgt][acgt][acgt][acgt]atc',
 868          'BsaXI' => 'ac[acgt][acgt][acgt][acgt][acgt]ctcc',
 869          'BsbI' => 'caacac',
 870          'BscGI' => 'cccgt',
 871          'BseMII' => 'ctcag',
 872          'BsePI' => 'gcgcgc',
 873          'BseRI' => 'gaggag',
 874          'BseSI' => 'g[gt]gc[ac]c',
 875          'BsgI' => 'gtgcag',
 876          'BsiI' => 'cacgag',
 877          'BsiYI' => 'cc[acgt][acgt][acgt][acgt][acgt][acgt][acgt]gg',
 878          'BsmAI' => 'gtctc',
 879          'BsmI' => 'gaatgc',
 880          'Bsp1407I' => 'tgtaca',
 881          'Bsp24I' => 'gac[acgt][acgt][acgt][acgt][acgt][acgt]tgg',
 882          'BspGI' => 'ctggac',
 883          'BspHI' => 'tcatga',
 884          'BspLU11I' => 'acatgt',
 885          'BspMI' => 'acctgc',
 886          'BspMII' => 'tccgga',
 887          'BsrBI' => 'ccgctc',
 888          'BsrDI' => 'gcaatg',
 889          'BsrI' => 'actgg',
 890          'BstEII' => 'ggt[acgt]acc',
 891          'BstXI' => 'cca[acgt][acgt][acgt][acgt][acgt][acgt]tgg',
 892          'BtrI' => 'cacgtc',
 893          'BtsI' => 'gcagtg',
 894          'Cac8I' => 'gc[acgt][acgt]gc',
 895          'CauII' => 'cc[cg]gg',
 896          'Cfr10I' => '[ag]ccgg[ct]',
 897          'CfrI' => '[ct]ggcc[ag]',
 898          'CjeI' => 'cca[acgt][acgt][acgt][acgt][acgt][acgt]gt',
 899          'CjePI' => 'cca[acgt][acgt][acgt][acgt][acgt][acgt][acgt]tc',
 900          'ClaI' => 'atcgat',
 901          'CviJI' => '[ag]gc[ct]',
 902          'CviRI' => 'tgca',
 903          'DdeI' => 'ct[acgt]ag',
 904          'DpnI' => 'gatc',
 905          'DraII' => '[ag]gg[acgt]cc[ct]',
 906          'DraIII' => 'cac[acgt][acgt][acgt]gtg',
 907          'DrdI' => 'gac[acgt][acgt][acgt][acgt][acgt][acgt]gtc',
 908          'DrdII' => 'gaacca',
 909          'DsaI' => 'cc[ag][ct]gg',
 910          'Eam1105I' => 'gac[acgt][acgt][acgt][acgt][acgt]gtc',
 911          'EciI' => 'ggcgga',
 912          'Eco31I' => 'ggtctc',
 913          'Eco47III' => 'agcgct',
 914          'Eco57I' => 'ctgaag',
 915          'EcoNI' => 'cct[acgt][acgt][acgt][acgt][acgt]agg',
 916          'EcoRI' => 'gaattc',
 917          'EcoRII' => 'cc[at]gg',
 918          'EcoRV' => 'gatatc',
 919          'Esp3I' => 'cgtctc',
 920          'EspI' => 'gct[acgt]agc',
 921          'FauI' => 'cccgc',
 922          'FinI' => 'gggac',
 923          'Fnu4HI' => 'gc[acgt]gc',
 924          'FnuDII' => 'cgcg',
 925          'FokI' => 'ggatg',
 926          'FseI' => 'ggccggcc',
 927          'GdiII' => 'cggcc[ag]',
 928          'GsuI' => 'ctggag',
 929          'HaeI' => '[at]ggcc[at]',
 930          'HaeII' => '[ag]gcgc[ct]',
 931          'HaeIII' => 'ggcc',
 932          'HaeIV' => 'ga[ct][acgt][acgt][acgt][acgt][acgt][ag]tc',
 933          'HgaI' => 'gacgc',
 934          'HgiAI' => 'g[at]gc[at]c',
 935          'HgiCI' => 'gg[ct][ag]cc',
 936          'HgiEII' => 'acc[acgt][acgt][acgt][acgt][acgt][acgt]ggt',
 937          'HgiJII' => 'g[ag]gc[ct]c',
 938          'HhaI' => 'gcgc',
 939          'Hin4I' => 'ga[cgt][acgt][acgt][acgt][acgt][acgt][acg]tc',
 940          'HindII' => 'gt[ct][ag]ac',
 941          'HindIII' => 'aagctt',
 942          'HinfI' => 'ga[acgt]tc',
 943          'HpaI' => 'gttaac',
 944          'HpaII' => 'ccgg',
 945          'HphI' => 'ggtga',
 946          'Hpy178III' => 'tc[acgt][acgt]ga',
 947          'Hpy188I' => 'tc[acgt]ga',
 948          'Hpy99I' => 'cg[at]cg',
 949          'KpnI' => 'ggtacc',
 950          'Ksp632I' => 'ctcttc',
 951          'MaeI' => 'ctag',
 952          'MaeII' => 'acgt',
 953          'MaeIII' => 'gt[acgt]ac',
 954          'MboI' => 'gatc',
 955          'MboII' => 'gaaga',
 956          'McrI' => 'cg[ag][ct]cg',
 957          'MfeI' => 'caattg',
 958          'MjaIV' => 'gt[acgt][acgt]ac',
 959          'MluI' => 'acgcgt',
 960          'MmeI' => 'tcc[ag]ac',
 961          'MnlI' => 'cctc',
 962          'MseI' => 'ttaa',
 963          'MslI' => 'ca[ct][acgt][acgt][acgt][acgt][ag]tg',
 964          'MstI' => 'tgcgca',
 965          'MwoI' => 'gc[acgt][acgt][acgt][acgt][acgt][acgt][acgt]gc',
 966          'NaeI' => 'gccggc',
 967          'NarI' => 'ggcgcc',
 968          'NcoI' => 'ccatgg',
 969          'NdeI' => 'catatg',
 970          'NheI' => 'gctagc',
 971          'NlaIII' => 'catg',
 972          'NlaIV' => 'gg[acgt][acgt]cc',
 973          'NotI' => 'gcggccgc',
 974          'NruI' => 'tcgcga',
 975          'NspBII' => 'c[ac]gc[gt]g',
 976          'NspI' => '[ag]catg[ct]',
 977          'PacI' => 'ttaattaa',
 978          'Pfl1108I' => 'tcgtag',
 979          'PflMI' => 'cca[acgt][acgt][acgt][acgt][acgt]tgg',
 980          'PleI' => 'gagtc',
 981          'PmaCI' => 'cacgtg',
 982          'PmeI' => 'gtttaaac',
 983          'PpiI' => 'gaac[acgt][acgt][acgt][acgt][acgt]ctc',
 984          'PpuMI' => '[ag]gg[at]cc[ct]',
 985          'PshAI' => 'gac[acgt][acgt][acgt][acgt]gtc',
 986          'PsiI' => 'ttataa',
 987          'PstI' => 'ctgcag',
 988          'PvuI' => 'cgatcg',
 989          'PvuII' => 'cagctg',
 990          'RleAI' => 'cccaca',
 991          'RsaI' => 'gtac',
 992          'RsrII' => 'cgg[at]ccg',
 993          'SacI' => 'gagctc',
 994          'SacII' => 'ccgcgg',
 995          'SalI' => 'gtcgac',
 996          'SanDI' => 'ggg[at]ccc',
 997          'SapI' => 'gctcttc',
 998          'SauI' => 'cct[acgt]agg',
 999          'ScaI' => 'agtact',
1000          'ScrFI' => 'cc[acgt]gg',
1001          'SduI' => 'g[agt]gc[act]c',
1002          'SecI' => 'cc[acgt][acgt]gg',
1003          'SexAI' => 'acc[at]ggt',
1004          'SfaNI' => 'gcatc',
1005          'SfeI' => 'ct[ag][ct]ag',
1006          'SfiI' => 'ggcc[acgt][acgt][acgt][acgt][acgt]ggcc',
1007          'SgfI' => 'gcgatcgc',
1008          'SgrAI' => 'c[ag]ccgg[ct]g',
1009          'SimI' => 'gggtc',
1010          'SmaI' => 'cccggg',
1011          'SmlI' => 'ct[ct][ag]ag',
1012          'SnaBI' => 'tacgta',
1013          'SnaI' => 'gtatac',
1014          'SpeI' => 'actagt',
1015          'SphI' => 'gcatgc',
1016          'SplI' => 'cgtacg',
1017          'SrfI' => 'gcccgggc',
1018          'Sse232I' => 'cgccggcg',
1019          'Sse8387I' => 'cctgcagg',
1020          'Sse8647I' => 'agg[at]cct',
1021          'SspI' => 'aatatt',
1022          'Sth132I' => 'cccg',
1023          'StuI' => 'aggcct',
1024          'StyI' => 'cc[at][at]gg',
1025          'SwaI' => 'atttaaat',
1026          'TaqI' => 'tcga',
1027          'TaqII' => 'gaccga',
1028          'TatI' => '[at]gtac[at]',
1029          'TauI' => 'gc[cg]gc',
1030          'TfiI' => 'ga[at]tc',
1031          'TseI' => 'gc[at]gc',
1032          'Tsp45I' => 'gt[cg]ac',
1033          'Tsp4CI' => 'ac[acgt]gt',
1034          'TspEI' => 'aatt',
1035          'TspRI' => 'ca[cg]tg[acgt][acgt]',
1036          'Tth111I' => 'gac[acgt][acgt][acgt]gtc',
1037          'Tth111II' => 'caa[ag]ca',
1038          'UbaGI' => 'cac[acgt][acgt][acgt][acgt]gtg',
1039          'UbaPI' => 'cgaacg',
1040          'VspI' => 'attaat',
1041          'XbaI' => 'tctaga',
1042          'XcmI' => 'cca[acgt][acgt][acgt][acgt][acgt][acgt][acgt][acgt][acgt]tgg',
1043          'XhoI' => 'ctcgag',
1044          'XhoII' => '[ag]gatc[ct]',
1045          'XmaIII' => 'cggccg',
1046          'XmnI' => 'gaa[acgt][acgt][acgt][acgt]ttc'
1047         );
1048
1049     return %enzymes;
1050 }
1051
1052 1;