Bio/Variation/VariantI.pm

   1 # $Id$
   2 #
   3 # BioPerl module for Bio::Variation::VariantI
   4 #
   5 # Cared for by Heikki Lehvaslaiho <heikki-at-bioperl-dot-org>
   6 #
   7 # Copyright Heikki Lehvaslaiho
   8 #
   9 # You may distribute this module under the same terms as perl itself
  10
  11 # POD documentation - main docs before the code
  12
  13 =head1 NAME
  14
  15 Bio::Variation::VariantI - Sequence Change SeqFeature abstract class
  16
  17 =head1 SYNOPSIS
  18
  19   #get Bio::Variant::VariantI somehow
  20   print $var->restriction_changes, "\n";
  21   foreach $allele ($var->each_Allele) {
  22       #work on Bio::Variation::Allele objects
  23   }
  24
  25 =head1 DESCRIPTION
  26
  27 This superclass defines common methods to basic sequence changes.  The
  28 instantiable classes Bio::Variation::DNAMutation,
  29 Bio::Variation::RNAChange and Bio::Variation::AAChange use them.
  30 See L<Bio::Variation::DNAMutation>, L<Bio::Variation::RNAChange>,
  31 and L<Bio::Variation::AAChange> for more information.
  32
  33 These classes store information, heavy computation to detemine allele
  34 sequences is done elsewhere.
  35
  36 The database cross-references are implemented as
  37 Bio::Annotation::DBLink objects. The methods to access them are
  38 defined in Bio::DBLinkContainerI. See L<Bio::Annotation::DBLink>
  39 and L<Bio::DBLinkContainerI> for details.
  40
  41 Bio::Variation::VariantI redifines and extends
  42 Bio::SeqFeature::Generic for sequence variations. This class
  43 describes specific sequence change events. These events are always
  44 from a specific reference sequence to something different. See
  45 L<Bio::SeqFeature::Generic> for more information.
  46
  47 IMPORTANT: The notion of reference sequence permeates all
  48 Bio::Variation classes. This is especially important to remember when
  49 dealing with Alleles. In a polymorphic site, there can be a large
  50 number of alleles. One of then has to be selected to be the reference
  51 allele (allele_ori). ALL the rest has to be passed to the Variant
  52 using the method add_Allele, including the mutated allele in a
  53 canonical mutation. The IO modules and generated attributes depend on
  54 it. They ignore the allele linked to using allele_mut and circulate
  55 each Allele returned by each_Allele into allele_mut and calculate
  56 the changes between that and allele_ori.
  57
  58
  59 =head1 FEEDBACK
  60
  61 =head2 Mailing Lists
  62
  63 User feedback is an integral part of the evolution of this and other
  64 Bioperl modules. Send your comments and suggestions preferably to the
  65 Bioperl mailing lists  Your participation is much appreciated.
  66
  67   bioperl-l@bioperl.org                  - General discussion
  68   http://bioperl.org/wiki/Mailing_lists  - About the mailing lists
  69
  70 =head2 Reporting Bugs
  71
  72 Report bugs to the Bioperl bug tracking system to help us keep track
  73 the bugs and their resolution.  Bug reports can be submitted via the
  74 web:
  75
  76   http://bugzilla.open-bio.org/
  77
  78 =head1 AUTHOR - Heikki Lehvaslaiho
  79
  80 Email:  heikki-at-bioperl-dot-org
  81
  82 =head1 APPENDIX
  83
  84 The rest of the documentation details each of the object
  85 methods. Internal methods are usually preceded with a _
  86
  87 =cut
  88
  89
  90 # Let the code begin...
  91
  92
  93 package Bio::Variation::VariantI;
  94 use strict;
  95 # Object preamble - inheritance
  96
  97 use base qw(Bio::Root::Root Bio::SeqFeature::Generic Bio::DBLinkContainerI);
  98
  99 =head2 id
 100
 101  Title   : id
 102  Usage   : $obj->id
 103  Function:
 104
 105            Read only method. Returns the id of the variation object.
 106            The id is the id of the first DBLink object attached to this object.
 107
 108  Example :
 109  Returns : scalar
 110  Args    : none
 111
 112 =cut
 113
 114 sub id {
 115    my ($self) = @_;
 116    my @ids = $self->each_DBLink;
 117    my $id = $ids[0] if scalar @ids > 0;
 118    return $id->database. "::". $id->primary_id if $id;
 119 }
 120
 121
 122 =head2 add_Allele
 123
 124  Title   : add_Allele
 125  Usage   : $self->add_Allele($allele)
 126  Function:
 127
 128             Adds one Bio::Variation::Allele into the list of alleles.
 129             Note that the method forces the convention that nucleotide
 130             sequence is in lower case and amino acds are in upper
 131             case.
 132
 133  Example :
 134  Returns : 1 when succeeds, 0 for failure.
 135  Args    : Allele object
 136
 137 =cut
 138
 139
 140 sub add_Allele {
 141   my ($self,$value) = @_;
 142   if (defined $value) {
 143       if( ! $value->isa('Bio::Variation::Allele') ) {
 144           my $com = ref $value;
 145           $self->throw("Is not a Allele object but a  [$com]");
 146           return 0;
 147       } else {
 148           if ( $self->isa('Bio::Variation::AAChange') ) {
 149               $value->seq( uc $value->seq) if $value->seq;
 150           } else {
 151               $value->seq( lc $value->seq) if $value->seq;
 152           }
 153           push(@{$self->{'alleles'}},$value);
 154           $self->allele_mut($value); #????
 155           return 1;
 156       }
 157   } else {
 158       return 0;
 159   }
 160 }
 161
 162
 163 =head2 each_Allele
 164
 165  Title   : alleles
 166  Usage   : $obj->each_Allele();
 167  Function:
 168
 169              Returns a list of Bio::Variation::Allele objects
 170
 171  Example :
 172  Returns : list of Alleles
 173  Args    : none
 174
 175 =cut
 176
 177 sub each_Allele{
 178    my ($self,@args) = @_;
 179    return @{$self->{'alleles'}};
 180 }
 181
 182
 183 =head2 isMutation
 184
 185  Title   : isMutation
 186  Usage   : print join('/', $obj->each_Allele) if not $obj->isMutation;
 187  Function:
 188
 189            Returns or sets the boolean value indicating that the
 190            variant descibed is a canonical mutation with two alleles
 191            assinged to be the original (wild type) allele and mutated
 192            allele, respectively. If this value is not set, it is
 193            assumed that the Variant descibes polymorphisms.
 194
 195  Returns : a boolean
 196
 197 =cut
 198
 199 sub isMutation {
 200     my ($self,$value) = @_;
 201     if (defined $value) {
 202         if ($value ) {
 203             $self->{'isMutation'} = 1;
 204         } else {
 205             $self->{'isMutation'} = 0;
 206         }
 207     }
 208     return $self->{'isMutation'};
 209 }
 210
 211
 212 =head2 allele_ori
 213
 214  Title   : allele_ori
 215  Usage   : $obj->allele_ori();
 216  Function:
 217
 218             Links to and returns the Bio::Variation::Allele object.
 219             If value is not set, returns false. All other Alleles are
 220             compared to this.
 221
 222             Amino acid sequences are stored in upper case characters,
 223             others in lower case.
 224
 225  Example :
 226  Returns : string
 227  Args    : string
 228
 229 See L<Bio::Variation::Allele> for more.
 230
 231 =cut
 232
 233 sub allele_ori {
 234    my ($self,$value) = @_;
 235    if( defined $value) {
 236        if ( ! ref $value || ! $value->isa('Bio::Variation::Allele')) {
 237            $self->throw("Value is not Bio::Variation::Allele but [$value]");
 238        } else {
 239            if ( $self->isa('Bio::Variation::AAChange') ) {
 240                $value->seq( uc $value->seq) if $value->seq;
 241            } else {
 242                $value->seq( lc $value->seq) if $value->seq;
 243            }
 244            $self->{'allele_ori'} = $value;
 245        }
 246    }
 247    return $self->{'allele_ori'};
 248 }
 249
 250
 251 =head2 allele_mut
 252
 253  Title   : allele_mut
 254  Usage   : $obj->allele_mut();
 255  Function:
 256
 257              Links to and returns the Bio::Variation::Allele
 258              object.  Sets and returns the mutated allele sequence.
 259              If value is not set, returns false.
 260
 261              Amino acid sequences are stored in upper case characters,
 262              others in lower case.
 263
 264  Example :
 265  Returns : string
 266  Args    : string
 267
 268 See L<Bio::Variation::Allele> for more.
 269
 270 =cut
 271
 272
 273 sub allele_mut {
 274    my ($self,$value) = @_;
 275    if( defined $value) {
 276        if ( ! ref $value || ! $value->isa('Bio::Variation::Allele')) {
 277            $self->throw("Value is not Bio::Variation::Allele but [$value]");
 278        } else {
 279            if ( $self->isa('Bio::Variation::AAChange') ) {
 280                $value->seq( uc $value->seq) if $value->seq;
 281            } else {
 282                $value->seq( lc $value->seq) if $value->seq;
 283            }
 284            $self->{'allele_mut'} = $value;
 285        }
 286    }
 287    return $self->{'allele_mut'};
 288 }
 289
 290 =head2 length
 291
 292  Title   : length
 293  Usage   : $obj->length();
 294  Function:
 295
 296             Sets and returns the length of the affected original
 297             allele sequence.  If value is not set, returns false == 0.
 298
 299             Value 0 means that the variant position is before the
 300             start=end sequence position. (Value 1 would denote a point
 301             mutation). This follows the convension to report an
 302             insertion (2insT) in equivalent way to a corresponding
 303             deletion (2delT) (Think about indel polymorpism ATC <=> AC
 304             where the origianal state is not known ).
 305
 306  Example :
 307  Returns : string
 308  Args    : string
 309
 310 =cut
 311
 312
 313 sub length {
 314    my ($self,$value) = @_;
 315    if ( defined $value) {
 316        $self->{'length'} = $value;
 317   }
 318    if ( ! exists $self->{'length'} ) {
 319        return 0;
 320    }
 321    return $self->{'length'};
 322 }
 323
 324 =head2 upStreamSeq
 325
 326  Title   : upStreamSeq
 327  Usage   : $obj->upStreamSeq();
 328  Function:
 329
 330             Sets and returns upstream flanking sequence string.  If
 331             value is not set, returns false. The sequence should be
 332             >=25 characters long, if possible.
 333
 334  Example :
 335  Returns : string or false
 336  Args    : string
 337
 338 =cut
 339
 340
 341 sub upStreamSeq {
 342     my ($self,$value) = @_;
 343     if( defined $value) {
 344         $self->{'upstreamseq'} = $value;
 345     }
 346    return $self->{'upstreamseq'};
 347 }
 348
 349
 350 =head2 dnStreamSeq
 351
 352  Title   : dnStreamSeq
 353  Usage   : $obj->dnStreamSeq();
 354  Function:
 355
 356             Sets and returns dnstream flanking sequence string.  If
 357             value is not set, returns false. The sequence should be
 358             >=25 characters long, if possible.
 359
 360  Example :
 361  Returns : string or false
 362  Args    : string
 363
 364 =cut
 365
 366
 367 sub dnStreamSeq {
 368     my ($self,$value) = @_;
 369     if( defined $value) {
 370         $self->{'dnstreamseq'} = $value;
 371     }
 372     return $self->{'dnstreamseq'};
 373
 374 }
 375
 376
 377 =head2 label
 378
 379  Title   : label
 380  Usage   : $obj->label();
 381  Function:
 382
 383             Sets and returns mutation event label(s).  If value is not
 384             set, or no argument is given returns false.  Each
 385             instantiable class needs to implement this method. Valid
 386             values are listed in 'Mutation event controlled vocabulary' in
 387             http://www.ebi.ac.uk/mutations/recommendations/mutevent.html.
 388
 389  Example :
 390  Returns : string
 391  Args    : string
 392
 393 =cut
 394
 395
 396 sub label {
 397     my ($self,$value) = @_;
 398     $self->throw_not_implemented();
 399 }
 400
 401
 402
 403 =head2 status
 404
 405  Title   : status
 406  Usage   : $obj->status()
 407  Function:
 408
 409            Returns the status of the sequence change object.
 410            Valid values are: 'suspected' and 'proven'
 411
 412  Example : $obj->status('proven');
 413  Returns : scalar
 414  Args    : valid string (optional, for setting)
 415
 416
 417 =cut
 418
 419
 420 sub status {
 421    my ($self,$value) = @_;
 422    my %status = (suspected => 1,
 423                  proven => 1
 424                  );
 425
 426    if( defined $value) {
 427        $value = lc $value;
 428        if ($status{$value}) {
 429            $self->{'status'} = $value;
 430        }
 431        else {
 432            $self->throw("$value is not valid status value!");
 433        }
 434     }
 435    if( ! exists $self->{'status'} ) {
 436        return "$self";
 437    }
 438    return $self->{'status'};
 439 }
 440
 441
 442 =head2 proof
 443
 444  Title   : proof
 445  Usage   : $obj->proof()
 446  Function:
 447
 448            Returns the proof of the sequence change object.
 449            Valid values are: 'computed' and 'experimental'.
 450
 451  Example : $obj->proof('computed');
 452  Returns : scalar
 453  Args    : valid string (optional, for setting)
 454
 455
 456 =cut
 457
 458
 459 sub proof {
 460     my ($self,$value) = @_;
 461     my %proof = (computed => 1,
 462                  experimental => 1
 463                  );
 464
 465     if( defined $value) {
 466         $value = lc $value;
 467         if ($proof{$value}) {
 468             $self->{'proof'} = $value;
 469         } else {
 470             $self->throw("$value is not valid proof value!");
 471         }
 472     }
 473     return $self->{'proof'};
 474 }
 475
 476
 477 =head2 region
 478
 479  Title   : region
 480  Usage   : $obj->region();
 481  Function:
 482
 483             Sets and returns the name of the sequence region type or
 484             protein domain at this location.  If value is not set,
 485             returns false.
 486
 487  Example :
 488  Returns : string
 489  Args    : string
 490
 491 =cut
 492
 493
 494 sub region {
 495     my ($self,$value) = @_;
 496     if( defined $value) {
 497         $self->{'region'} = $value;
 498     }
 499     return $self->{'region'};
 500 }
 501
 502
 503 =head2 region_value
 504
 505  Title   : region_value
 506  Usage   : $obj->region_value();
 507  Function:
 508
 509             Sets and returns the name of the sequence region_value or
 510             protein domain at this location.  If value is not set,
 511             returns false.
 512
 513  Example :
 514  Returns : string
 515  Args    : string
 516
 517 =cut
 518
 519
 520 sub region_value {
 521     my ($self,$value) = @_;
 522     if( defined $value) {
 523         $self->{'region_value'} = $value;
 524     }
 525     return $self->{'region_value'};
 526 }
 527
 528 =head2 region_dist
 529
 530  Title   : region_dist
 531  Usage   : $obj->region_dist();
 532  Function:
 533
 534             Sets and returns the distance tot the closest region
 535             (i.e. intro/exon or domain) boundary. If distance is not
 536             set, returns false.
 537
 538  Example :
 539  Returns : integer
 540  Args    : integer
 541
 542 =cut
 543
 544
 545 sub region_dist {
 546     my ($self,$value) = @_;
 547     if( defined $value) {
 548        if (  not $value =~ /^[+-]?\d+$/ ) {
 549            $self->throw("[$value] for region_dist has to be an integer\n");
 550         } else {
 551             $self->{'region_dist'} = $value;
 552         }
 553     }
 554     return $self->{'region_dist'};
 555 }
 556
 557
 558 =head2 numbering
 559
 560  Title   : numbering
 561  Usage   : $obj->numbering()
 562  Function:
 563
 564            Returns the numbering chema used locating sequnce features.
 565            Valid values are: 'entry' and 'coding'
 566
 567  Example : $obj->numbering('coding');
 568  Returns : scalar
 569  Args    : valid string (optional, for setting)
 570
 571
 572 =cut
 573
 574
 575 sub numbering {
 576    my ($self,$value) = @_;
 577    my %numbering = (entry => 1,
 578                     coding => 1
 579                     );
 580
 581    if( defined $value) {
 582        $value = lc $value;
 583        if ($numbering{$value}) {
 584            $self->{'numbering'} = $value;
 585        }
 586        else {
 587            $self->throw("'$value' is not a valid for numbering!");
 588        }
 589     }
 590    if( ! exists $self->{'numbering'} ) {
 591        return "$self";
 592    }
 593    return $self->{'numbering'};
 594 }
 595
 596 =head2 mut_number
 597
 598  Title   : mut_number
 599  Usage   : $num = $obj->mut_number;
 600          : $num = $obj->mut_number($number);
 601  Function:
 602
 603            Returns or sets the number identifying the order in which the
 604            mutation has been issued. Numbers shouldstart from 1.
 605            If the number has never been set, the method will return ''
 606
 607            If you want the output from IO modules look nice and, for
 608            multivariant/allele variations, make sense you better set
 609            this attribute.
 610
 611  Returns : an integer
 612
 613 =cut
 614
 615
 616 sub mut_number {
 617     my ($self,$value) = @_;
 618     if (defined $value) {
 619         $self->{'mut_number'} = $value;
 620     }
 621     unless (exists $self->{'mut_number'}) {
 622         return ('');
 623     } else {
 624         return $self->{'mut_number'};
 625     }
 626 }
 627
 628
 629 =head2 SeqDiff
 630
 631  Title   : SeqDiff
 632  Usage   : $mutobj = $obj->SeqDiff;
 633          : $mutobj = $obj->SeqDiff($objref);
 634  Function:
 635
 636            Returns or sets the link-reference to the umbrella
 637            Bio::Variation::SeqDiff object.  If there is no link,
 638            it will return undef
 639
 640            Note: Adding a variant into a SeqDiff object will
 641            automatically set this value.
 642
 643  Returns : an obj_ref or undef
 644
 645 See L<Bio::Variation::SeqDiff> for more information.
 646
 647 =cut
 648
 649 sub SeqDiff {
 650     my ($self,$value) = @_;
 651     if (defined $value) {
 652         if( ! $value->isa('Bio::Variation::SeqDiff') ) {
 653             $self->throw("Is not a Bio::Variation::SeqDiff object but a [$value]");
 654             return;
 655         }
 656         else {
 657             $self->{'seqDiff'} = $value;
 658         }
 659     }
 660     unless (exists $self->{'seqDiff'}) {
 661         return;
 662     } else {
 663         return $self->{'seqDiff'};
 664     }
 665 }
 666
 667 =head2 add_DBLink
 668
 669  Title   : add_DBLink
 670  Usage   : $self->add_DBLink($ref)
 671  Function: adds a link object
 672  Example :
 673  Returns :
 674  Args    :
 675
 676
 677 =cut
 678
 679
 680 sub add_DBLink{
 681    my ($self,$com) = @_;
 682    if( $com && ! $com->isa('Bio::Annotation::DBLink') ) {
 683        $self->throw("Is not a link object but a  [$com]");
 684    }
 685    $com && push(@{$self->{'link'}},$com);
 686 }
 687
 688 =head2 each_DBLink
 689
 690  Title   : each_DBLink
 691  Usage   : foreach $ref ( $self->each_DBlink() )
 692  Function: gets an array of DBlink of objects
 693  Example :
 694  Returns :
 695  Args    :
 696
 697
 698 =cut
 699
 700 sub each_DBLink{
 701    my ($self) = @_;
 702
 703    return @{$self->{'link'}};
 704 }
 705
 706 =head2 restriction_changes
 707
 708  Title   : restriction_changes
 709  Usage   : $obj->restriction_changes();
 710  Function:
 711
 712             Returns a string containing a list of restriction
 713             enzyme changes of form +EcoRI, separated by
 714             commas. Strings need to be valid restriction enzyme names
 715             as stored in REBASE. allele_ori and allele_mut need to be assigned.
 716
 717  Example :
 718  Returns : string
 719  Args    : string
 720
 721 =cut
 722
 723 sub restriction_changes {
 724     my ($self) = @_;
 725
 726     if (not $self->{'re_changes'}) {
 727         my %re = &_enzymes;
 728
 729         # complain if used on AA data
 730         if ($self->isa('Bio::Variation::AAChange')) {
 731             $self->throw('Restriction enzymes do not bite polypeptides!');
 732         }
 733
 734         #sanity checks
 735         $self->warn('Upstream sequence is empty!')
 736             if $self->upStreamSeq eq '';
 737         $self->warn('Downstream sequence is empty!')
 738             if $self->dnStreamSeq eq '';
 739 #        $self->warn('Original allele sequence is empty!')
 740 #            if $self->allele_ori eq '';
 741 #        $self->warn('Mutated allele sequence is empty!')
 742 #            if $self->allele_mut eq '';
 743
 744         #reuse the non empty DNA level list at RNA level if the flanks are identical
 745         #Hint: Check DNAMutation object first
 746         if ($self->isa('Bio::Variation::RNAChange') and  $self->DNAMutation and
 747             $self->upStreamSeq eq $self->DNAMutation->upStreamSeq  and
 748             $self->dnStreamSeq eq $self->DNAMutation->dnStreamSeq and
 749             $self->DNAMutation->restriction_changes ne '' ) {
 750             $self->{'re_changes'} = $self->DNAMutation->restriction_changes;
 751         } else {
 752
 753             #maximum length of a type II restriction site in the current REBASE
 754             my ($le_dn) = 15;
 755             my ($le_up) = $le_dn;
 756
 757             #reduce the flank lengths if the desired length is not available
 758             $le_dn = CORE::length ($self->dnStreamSeq) if $le_dn > CORE::length ($self->dnStreamSeq);
 759             $le_up = CORE::length ($self->upStreamSeq) if $le_up > CORE::length ($self->upStreamSeq);
 760
 761             #Build sequence strings to compare
 762             my ($oriseq, $mutseq);
 763             $oriseq  = $mutseq = substr($self->upStreamSeq, -$le_up, $le_up);
 764             $oriseq .= $self->allele_ori->seq if $self->allele_ori->seq;
 765             $mutseq .= $self->allele_mut->seq if $self->allele_mut->seq;
 766             $oriseq .= substr($self->dnStreamSeq, 0, $le_dn);
 767             $mutseq .= substr($self->dnStreamSeq, 0, $le_dn);
 768
 769             # ... and their reverse complements
 770             my $oriseq_rev = _revcompl ($oriseq);
 771             my $mutseq_rev = _revcompl ($mutseq);
 772
 773             # collect results into a string
 774             my $rec = '';
 775             foreach my $enz (sort keys (%re)) {
 776                 my $site = $re{$enz};
 777                 my @ori = ($oriseq=~ /$site/g);
 778                 my @mut = ($mutseq=~ /$site/g);
 779                 my @ori_r = ($oriseq_rev =~ /$site/g);
 780                 my @mut_r = ($mutseq_rev =~ /$site/g);
 781
 782                 $rec .= '+'. $enz. ", "
 783                     if (scalar @ori < scalar @mut) or (scalar @ori_r < scalar @mut_r);
 784                 $rec .= '-'. $enz. ", "
 785                     if (scalar @ori > scalar @mut) or (scalar @ori_r > scalar @mut_r);
 786
 787             }
 788             $rec = substr($rec, 0, CORE::length($rec) - 2) if $rec ne '';
 789             $self->{'re_changes'} =  $rec;
 790         }
 791     }
 792     return $self->{'re_changes'}
 793 }
 794
 795
 796 sub _revcompl {
 797     # side effect: lower case letters
 798     my ($seq) = shift;
 799
 800     $seq = lc $seq;
 801     $seq =~ tr/acgtrymkswhbvdnx/tgcayrkmswdvbhnx/;
 802     return CORE::reverse $seq;
 803 }
 804
 805
 806 sub _enzymes {
 807  #REBASE version 005   type2.005
 808  my %enzymes =  (
 809          'AarI' => 'cacctgc',
 810          'AatII' => 'gacgtc',
 811          'AccI' => 'gt[ac][gt]ac',
 812          'AceIII' => 'cagctc',
 813          'AciI' => 'ccgc',
 814          'AclI' => 'aacgtt',
 815          'AcyI' => 'g[ag]cg[ct]c',
 816          'AflII' => 'cttaag',
 817          'AflIII' => 'ac[ag][ct]gt',
 818          'AgeI' => 'accggt',
 819          'AhaIII' => 'tttaaa',
 820          'AloI' => 'gaac[acgt][acgt][acgt][acgt][acgt][acgt]tcc',
 821          'AluI' => 'agct',
 822          'AlwNI' => 'cag[acgt][acgt][acgt]ctg',
 823          'ApaBI' => 'gca[acgt][acgt][acgt][acgt][acgt]tgc',
 824          'ApaI' => 'gggccc',
 825          'ApaLI' => 'gtgcac',
 826          'ApoI' => '[ag]aatt[ct]',
 827          'AscI' => 'ggcgcgcc',
 828          'AsuI' => 'gg[acgt]cc',
 829          'AsuII' => 'ttcgaa',
 830          'AvaI' => 'c[ct]cg[ag]g',
 831          'AvaII' => 'gg[at]cc',
 832          'AvaIII' => 'atgcat',
 833          'AvrII' => 'cctagg',
 834          'BaeI' => 'ac[acgt][acgt][acgt][acgt]gta[ct]c',
 835          'BalI' => 'tggcca',
 836          'BamHI' => 'ggatcc',
 837          'BbvCI' => 'cctcagc',
 838          'BbvI' => 'gcagc',
 839          'BbvII' => 'gaagac',
 840          'BccI' => 'ccatc',
 841          'Bce83I' => 'cttgag',
 842          'BcefI' => 'acggc',
 843          'BcgI' => 'cga[acgt][acgt][acgt][acgt][acgt][acgt]tgc',
 844          'BciVI' => 'gtatcc',
 845          'BclI' => 'tgatca',
 846          'BetI' => '[at]ccgg[at]',
 847          'BfiI' => 'actggg',
 848          'BglI' => 'gcc[acgt][acgt][acgt][acgt][acgt]ggc',
 849          'BglII' => 'agatct',
 850          'BinI' => 'ggatc',
 851          'BmgI' => 'g[gt]gccc',
 852          'BplI' => 'gag[acgt][acgt][acgt][acgt][acgt]ctc',
 853          'Bpu10I' => 'cct[acgt]agc',
 854          'BsaAI' => '[ct]acgt[ag]',
 855          'BsaBI' => 'gat[acgt][acgt][acgt][acgt]atc',
 856          'BsaXI' => 'ac[acgt][acgt][acgt][acgt][acgt]ctcc',
 857          'BsbI' => 'caacac',
 858          'BscGI' => 'cccgt',
 859          'BseMII' => 'ctcag',
 860          'BsePI' => 'gcgcgc',
 861          'BseRI' => 'gaggag',
 862          'BseSI' => 'g[gt]gc[ac]c',
 863          'BsgI' => 'gtgcag',
 864          'BsiI' => 'cacgag',
 865          'BsiYI' => 'cc[acgt][acgt][acgt][acgt][acgt][acgt][acgt]gg',
 866          'BsmAI' => 'gtctc',
 867          'BsmI' => 'gaatgc',
 868          'Bsp1407I' => 'tgtaca',
 869          'Bsp24I' => 'gac[acgt][acgt][acgt][acgt][acgt][acgt]tgg',
 870          'BspGI' => 'ctggac',
 871          'BspHI' => 'tcatga',
 872          'BspLU11I' => 'acatgt',
 873          'BspMI' => 'acctgc',
 874          'BspMII' => 'tccgga',
 875          'BsrBI' => 'ccgctc',
 876          'BsrDI' => 'gcaatg',
 877          'BsrI' => 'actgg',
 878          'BstEII' => 'ggt[acgt]acc',
 879          'BstXI' => 'cca[acgt][acgt][acgt][acgt][acgt][acgt]tgg',
 880          'BtrI' => 'cacgtc',
 881          'BtsI' => 'gcagtg',
 882          'Cac8I' => 'gc[acgt][acgt]gc',
 883          'CauII' => 'cc[cg]gg',
 884          'Cfr10I' => '[ag]ccgg[ct]',
 885          'CfrI' => '[ct]ggcc[ag]',
 886          'CjeI' => 'cca[acgt][acgt][acgt][acgt][acgt][acgt]gt',
 887          'CjePI' => 'cca[acgt][acgt][acgt][acgt][acgt][acgt][acgt]tc',
 888          'ClaI' => 'atcgat',
 889          'CviJI' => '[ag]gc[ct]',
 890          'CviRI' => 'tgca',
 891          'DdeI' => 'ct[acgt]ag',
 892          'DpnI' => 'gatc',
 893          'DraII' => '[ag]gg[acgt]cc[ct]',
 894          'DraIII' => 'cac[acgt][acgt][acgt]gtg',
 895          'DrdI' => 'gac[acgt][acgt][acgt][acgt][acgt][acgt]gtc',
 896          'DrdII' => 'gaacca',
 897          'DsaI' => 'cc[ag][ct]gg',
 898          'Eam1105I' => 'gac[acgt][acgt][acgt][acgt][acgt]gtc',
 899          'EciI' => 'ggcgga',
 900          'Eco31I' => 'ggtctc',
 901          'Eco47III' => 'agcgct',
 902          'Eco57I' => 'ctgaag',
 903          'EcoNI' => 'cct[acgt][acgt][acgt][acgt][acgt]agg',
 904          'EcoRI' => 'gaattc',
 905          'EcoRII' => 'cc[at]gg',
 906          'EcoRV' => 'gatatc',
 907          'Esp3I' => 'cgtctc',
 908          'EspI' => 'gct[acgt]agc',
 909          'FauI' => 'cccgc',
 910          'FinI' => 'gggac',
 911          'Fnu4HI' => 'gc[acgt]gc',
 912          'FnuDII' => 'cgcg',
 913          'FokI' => 'ggatg',
 914          'FseI' => 'ggccggcc',
 915          'GdiII' => 'cggcc[ag]',
 916          'GsuI' => 'ctggag',
 917          'HaeI' => '[at]ggcc[at]',
 918          'HaeII' => '[ag]gcgc[ct]',
 919          'HaeIII' => 'ggcc',
 920          'HaeIV' => 'ga[ct][acgt][acgt][acgt][acgt][acgt][ag]tc',
 921          'HgaI' => 'gacgc',
 922          'HgiAI' => 'g[at]gc[at]c',
 923          'HgiCI' => 'gg[ct][ag]cc',
 924          'HgiEII' => 'acc[acgt][acgt][acgt][acgt][acgt][acgt]ggt',
 925          'HgiJII' => 'g[ag]gc[ct]c',
 926          'HhaI' => 'gcgc',
 927          'Hin4I' => 'ga[cgt][acgt][acgt][acgt][acgt][acgt][acg]tc',
 928          'HindII' => 'gt[ct][ag]ac',
 929          'HindIII' => 'aagctt',
 930          'HinfI' => 'ga[acgt]tc',
 931          'HpaI' => 'gttaac',
 932          'HpaII' => 'ccgg',
 933          'HphI' => 'ggtga',
 934          'Hpy178III' => 'tc[acgt][acgt]ga',
 935          'Hpy188I' => 'tc[acgt]ga',
 936          'Hpy99I' => 'cg[at]cg',
 937          'KpnI' => 'ggtacc',
 938          'Ksp632I' => 'ctcttc',
 939          'MaeI' => 'ctag',
 940          'MaeII' => 'acgt',
 941          'MaeIII' => 'gt[acgt]ac',
 942          'MboI' => 'gatc',
 943          'MboII' => 'gaaga',
 944          'McrI' => 'cg[ag][ct]cg',
 945          'MfeI' => 'caattg',
 946          'MjaIV' => 'gt[acgt][acgt]ac',
 947          'MluI' => 'acgcgt',
 948          'MmeI' => 'tcc[ag]ac',
 949          'MnlI' => 'cctc',
 950          'MseI' => 'ttaa',
 951          'MslI' => 'ca[ct][acgt][acgt][acgt][acgt][ag]tg',
 952          'MstI' => 'tgcgca',
 953          'MwoI' => 'gc[acgt][acgt][acgt][acgt][acgt][acgt][acgt]gc',
 954          'NaeI' => 'gccggc',
 955          'NarI' => 'ggcgcc',
 956          'NcoI' => 'ccatgg',
 957          'NdeI' => 'catatg',
 958          'NheI' => 'gctagc',
 959          'NlaIII' => 'catg',
 960          'NlaIV' => 'gg[acgt][acgt]cc',
 961          'NotI' => 'gcggccgc',
 962          'NruI' => 'tcgcga',
 963          'NspBII' => 'c[ac]gc[gt]g',
 964          'NspI' => '[ag]catg[ct]',
 965          'PacI' => 'ttaattaa',
 966          'Pfl1108I' => 'tcgtag',
 967          'PflMI' => 'cca[acgt][acgt][acgt][acgt][acgt]tgg',
 968          'PleI' => 'gagtc',
 969          'PmaCI' => 'cacgtg',
 970          'PmeI' => 'gtttaaac',
 971          'PpiI' => 'gaac[acgt][acgt][acgt][acgt][acgt]ctc',
 972          'PpuMI' => '[ag]gg[at]cc[ct]',
 973          'PshAI' => 'gac[acgt][acgt][acgt][acgt]gtc',
 974          'PsiI' => 'ttataa',
 975          'PstI' => 'ctgcag',
 976          'PvuI' => 'cgatcg',
 977          'PvuII' => 'cagctg',
 978          'RleAI' => 'cccaca',
 979          'RsaI' => 'gtac',
 980          'RsrII' => 'cgg[at]ccg',
 981          'SacI' => 'gagctc',
 982          'SacII' => 'ccgcgg',
 983          'SalI' => 'gtcgac',
 984          'SanDI' => 'ggg[at]ccc',
 985          'SapI' => 'gctcttc',
 986          'SauI' => 'cct[acgt]agg',
 987          'ScaI' => 'agtact',
 988          'ScrFI' => 'cc[acgt]gg',
 989          'SduI' => 'g[agt]gc[act]c',
 990          'SecI' => 'cc[acgt][acgt]gg',
 991          'SexAI' => 'acc[at]ggt',
 992          'SfaNI' => 'gcatc',
 993          'SfeI' => 'ct[ag][ct]ag',
 994          'SfiI' => 'ggcc[acgt][acgt][acgt][acgt][acgt]ggcc',
 995          'SgfI' => 'gcgatcgc',
 996          'SgrAI' => 'c[ag]ccgg[ct]g',
 997          'SimI' => 'gggtc',
 998          'SmaI' => 'cccggg',
 999          'SmlI' => 'ct[ct][ag]ag',
1000          'SnaBI' => 'tacgta',
1001          'SnaI' => 'gtatac',
1002          'SpeI' => 'actagt',
1003          'SphI' => 'gcatgc',
1004          'SplI' => 'cgtacg',
1005          'SrfI' => 'gcccgggc',
1006          'Sse232I' => 'cgccggcg',
1007          'Sse8387I' => 'cctgcagg',
1008          'Sse8647I' => 'agg[at]cct',
1009          'SspI' => 'aatatt',
1010          'Sth132I' => 'cccg',
1011          'StuI' => 'aggcct',
1012          'StyI' => 'cc[at][at]gg',
1013          'SwaI' => 'atttaaat',
1014          'TaqI' => 'tcga',
1015          'TaqII' => 'gaccga',
1016          'TatI' => '[at]gtac[at]',
1017          'TauI' => 'gc[cg]gc',
1018          'TfiI' => 'ga[at]tc',
1019          'TseI' => 'gc[at]gc',
1020          'Tsp45I' => 'gt[cg]ac',
1021          'Tsp4CI' => 'ac[acgt]gt',
1022          'TspEI' => 'aatt',
1023          'TspRI' => 'ca[cg]tg[acgt][acgt]',
1024          'Tth111I' => 'gac[acgt][acgt][acgt]gtc',
1025          'Tth111II' => 'caa[ag]ca',
1026          'UbaGI' => 'cac[acgt][acgt][acgt][acgt]gtg',
1027          'UbaPI' => 'cgaacg',
1028          'VspI' => 'attaat',
1029          'XbaI' => 'tctaga',
1030          'XcmI' => 'cca[acgt][acgt][acgt][acgt][acgt][acgt][acgt][acgt][acgt]tgg',
1031          'XhoI' => 'ctcgag',
1032          'XhoII' => '[ag]gatc[ct]',
1033          'XmaIII' => 'cggccg',
1034          'XmnI' => 'gaa[acgt][acgt][acgt][acgt]ttc'
1035         );
1036
1037     return %enzymes;
1038 }
1039
1040 1;