branch-1-6/Bio/Tools/Protparam.pm

   1 # $Id$
   2 #
   3 # BioPerl module for Bio::Tools::Protparam
   4 #
   5 # Please direct questions and support issues to <bioperl-l@bioperl.org>
   6 #
   7 # Cared for by Richard Dobson, r.j.dobson at qmul dot ac dot uk
   8 #
   9 # Copyright Richard Dobson
  10 #
  11 # You may distribute this module under the same terms as perl itself
  12
  13 # POD documentation - main docs before the code
  14
  15 =head1 NAME
  16
  17 Bio::Tools::Protparam - submit to and parse output from protparam ;
  18
  19 =head1 SYNOPSIS
  20
  21   my $gb = new Bio::DB::GenBank(-retrievaltype => 'tempfile' ,
  22                                 -format => 'Fasta');
  23   my @ids=qw(O14521 O43709 O43826);
  24   my $seqio = $gb->get_Stream_by_acc(\@ids );
  25
  26   while( my $seq =  $seqio->next_seq ) {
  27
  28         my $pp = Protparam->new(seq=>$seq->seq);
  29
  30         print
  31         "ID : ", $seq->display_id,"\n",
  32         "Amino acid number : ",$pp->amino_acid_number(),"\n",
  33         "Number of negative amino acids : ",$pp->num_neg(),"\n",
  34         "Number of positive amino acids : ",$pp->num_pos(),"\n",
  35         "Molecular weight : ",$pp->molecular_weight(),"\n",
  36         "Theoretical pI : ",$pp->theoretical_pI(),"\n",
  37         "Total number of atoms : ", $pp->total_atoms(),"\n",
  38         "Number of carbon atoms : ",$pp->num_carbon(),"\n",
  39         "Number of hydrogen atoms : ",$pp->num_hydrogen(),"\n",
  40         "Number of nitrogen atoms : ",$pp->num_nitro(),"\n",
  41         "Number of oxygen atoms : ",$pp->num_oxygen(),"\n",
  42         "Number of sulphur atoms : ",$pp->num_sulphur(),"\n",
  43         "Half life : ", $pp->half_life(),"\n",
  44         "Instability Index : ", $pp->instability_index(),"\n",
  45         "Stability class : ", $pp->stability(),"\n",
  46         "Aliphatic_index : ",$pp->aliphatic_index(),"\n",
  47         "Gravy : ", $pp->gravy(),"\n",
  48         "Composition of A : ", $pp->AA_comp('A'),"\n",
  49         "Composition of R : ", $pp->AA_comp('R'),"\n",
  50         "Composition of N : ", $pp->AA_comp('N'),"\n",
  51         "Composition of D : ", $pp->AA_comp('D'),"\n",
  52         "Composition of C : ", $pp->AA_comp('C'),"\n",
  53         "Composition of Q : ", $pp->AA_comp('Q'),"\n",
  54         "Composition of E : ", $pp->AA_comp('E'),"\n",
  55         "Composition of G : ", $pp->AA_comp('G'),"\n",
  56         "Composition of H : ", $pp->AA_comp('H'),"\n",
  57         "Composition of I : ", $pp->AA_comp('I'),"\n",
  58         "Composition of L : ", $pp->AA_comp('L'),"\n",
  59         "Composition of K : ", $pp->AA_comp('K'),"\n",
  60         "Composition of M : ", $pp->AA_comp('M'),"\n",
  61         "Composition of F : ", $pp->AA_comp('F'),"\n",
  62         "Composition of P : ", $pp->AA_comp('P'),"\n",
  63         "Composition of S : ", $pp->AA_comp('S'),"\n",
  64         "Composition of T : ", $pp->AA_comp('T'),"\n",
  65         "Composition of W : ", $pp->AA_comp('W'),"\n",
  66         "Composition of Y : ", $pp->AA_comp('Y'),"\n",
  67         "Composition of V : ", $pp->AA_comp('V'),"\n",
  68         "Composition of B : ", $pp->AA_comp('B'),"\n",
  69         "Composition of Z : ", $pp->AA_comp('Z'),"\n",
  70         "Composition of X : ", $pp->AA_comp('X'),"\n";
  71 }
  72
  73 =head1 DESCRIPTION
  74
  75 This module takes an amino acid sequence and submits it to the
  76 Protparam program at www.expasy.org/cgi-bin/protparam.  Many
  77 properties of the submitted sequence are returned.
  78
  79 =head1 AUTHOR
  80
  81 Richard Dobson, r.j.dobson at qmul dot ac dot uk
  82
  83 =cut
  84
  85 # Let the code begin...
  86
  87 package Bio::Tools::Protparam;
  88
  89 use strict;
  90 use base qw(Bio::Root::Root);
  91 use LWP 5.64;
  92
  93 =head2 new
  94
  95   Title    : new
  96   Usage    : $pp = Protparam->new(seq=>$seq->seq);
  97   Function : Creates a new Protparam object
  98   Returns  : A Protparam object
  99   Args     : A sequence
 100
 101 =cut
 102
 103
 104
 105 sub new {
 106
 107
 108         my ($class,@args) = @_;
 109
 110
 111         @args=('-url'=>'http://www.expasy.org/cgi-bin/protparam','-form'=>'sequence',@args);
 112
 113         my $self=$class->SUPER::new(@args);
 114
 115         my ($url,$seq,$form)=$self->_rearrange([qw(URL SEQ FORM)],@args);
 116
 117         my $browser = LWP::UserAgent->new;
 118         my $response;
 119
 120         #send request to PROTPARAM @ Expasy
 121         $response = $browser->post($url,
 122                            [
 123                             $form => $seq
 124                            ],
 125                            'User-Agent' => 'Mozilla/4.76 [en] (Win2000; U)',
 126                           );
 127
 128         #Check if successful
 129         $self->throw("$self->{'url'} error: ", $response->status_line) unless $response->is_success;
 130         $self->throw("Bad content type at $self->{'url'} ", $response->content_type) unless $response->content_type eq 'text/html';
 131
 132         my $protParamOutput=$response->decoded_content;
 133
 134         $self->{'output'}=$protParamOutput;
 135
 136         return bless $self,$class;
 137
 138 }
 139
 140 =head2 num_neg
 141
 142   Title    : num_neg
 143   Usage    : $pp->num_neg()
 144   Function : Retrieves the number of negative amino acids in a sequence
 145   Returns  : Returns the number of negative amino acids in a sequence
 146   Args     : none
 147
 148 =cut
 149
 150
 151
 152 sub num_neg{
 153
 154         my $self=shift;
 155
 156         ($self->{'negAA'})=$self->{'output'}=~/<B>Total number of negatively charged residues.*?<\/B>\s*(\d*)/;
 157
 158         return $self->{'negAA'};
 159
 160
 161 }
 162
 163 =head2 num_pos
 164
 165   Title    : num_pos
 166   Usage    : $pp->num_pos()
 167   Function : Retrieves the number of positive amino acids in a sequence
 168   Returns  : Returns the number of positive amino acids in a sequence
 169   Args     : none
 170
 171 =cut
 172
 173
 174 sub num_pos{
 175
 176         my $self=shift;
 177
 178         ($self->{'posAA'})=$self->{'output'}=~/<B>Total number of positively charged residues.*?<\/B>\s*(\d*)/;
 179
 180         return $self->{'posAA'};
 181
 182
 183 }
 184
 185 =head2 amino_acid_number
 186
 187   Title    : amino_acid_number
 188   Usage    : $pp->amino_acid_number()
 189   Function : Retrieves the number of amino acids within a sequence
 190   Returns  : Returns the number of amino acids within a sequence
 191   Args     : none
 192
 193 =cut
 194
 195 sub amino_acid_number{
 196         my $self=shift;
 197
 198         ($self->{'numAA'})=$self->{'output'}=~/<B>Number of amino acids:<\/B> (\d+)/;
 199
 200         return $self->{'numAA'};
 201 }
 202
 203 =head2 total_atoms
 204
 205   Title    : total_atoms
 206   Usage    : $pp->total_atoms()
 207   Function : Retrieves the total number of atoms within a sequence
 208   Returns  : Returns the total number of atoms within a sequence
 209   Args     : none
 210
 211 =cut
 212
 213
 214 sub total_atoms{
 215         my $self=shift;
 216
 217         $self->{'total_atoms'}=$self->{'output'}=~/<B>Total number of atoms:<\/B>\s*(\d*)/;
 218
 219         return $self->{'total_atoms'};
 220 }
 221
 222 =head2 molecular_weight
 223
 224   Title    : molecular_weight
 225   Usage    : $pp->molecular_weight()
 226   Function : Retrieves the molecular weight of a sequence
 227   Returns  : Returns the molecular weight of a sequence
 228   Args     : none
 229
 230 =cut
 231
 232
 233 sub molecular_weight{
 234         my $self=shift;
 235         ($self->{'MolWt'})=$self->{'output'}=~/<B>Molecular weight:<\/B> (\d*\.{0,1}\d*)/;
 236         return $self->{'MolWt'};
 237
 238 }
 239
 240 =head2 theoretical_pI
 241
 242   Title    : theoretical_pI
 243   Usage    : $pp->theoretical_pI()
 244   Function : Retrieve the theoretical pI for a sequence
 245   Returns  : Return the theoretical pI for a sequence
 246   Args     : none
 247
 248 =cut
 249
 250
 251 sub theoretical_pI{
 252         my $self=shift;
 253         ($self->{'TpI'})=$self->{'output'}=~/<B>Theoretical pI:<\/B> (-{0,1}\d*\.{0,1}\d*)/;
 254         return $self->{'TpI'};
 255 }
 256
 257 =head2 num_carbon
 258
 259   Title    : num_carbon
 260   Usage    : $pp->num_carbon()
 261   Function : Retrieves the number of carbon atoms in a sequence
 262   Returns  : Returns the number of carbon atoms in a sequence
 263   Args     : none
 264
 265 =cut
 266
 267
 268 sub num_carbon{
 269         my $self=shift;
 270
 271         ($self->{'car'}) = $self->{'output'}=~/Carbon\s+C\s+(\d+)/;
 272
 273         return $self->{'car'};
 274
 275 }
 276
 277 =head2 num_hydrogen
 278
 279   Title    : num_hydrogen
 280   Usage    : $pp->num_hydrogen
 281   Function : Retrieves the number of hydrogen atoms in a sequence
 282   Returns  : Returns the number of hydrogen atoms in a sequence
 283   Args     : none
 284
 285 =cut
 286
 287
 288 sub num_hydrogen{
 289         my $self=shift;
 290
 291         ($self->{'hyd'}) = $self->{'output'}=~/Hydrogen\s+H\s+(\d+)/;
 292
 293         return $self->{'hyd'}
 294
 295 }
 296
 297 =head2 num_nitro
 298
 299   Title    : num_nitro
 300   Usage    : $pp->num_nitro
 301   Function : Retrieves the number of nitrogen atoms in a sequence
 302   Returns  : Returns the number of nitrogen atoms in a sequence
 303   Args     : none
 304
 305 =cut
 306
 307
 308 sub num_nitro{
 309         my $self=shift;
 310
 311         ($self->{'nitro'}) = $self->{'output'}=~/Nitrogen\s+N\s+(\d+)/;
 312
 313
 314         return $self->{'nitro'};
 315
 316 }
 317
 318 =head2 num_oxygen
 319
 320   Title    : num_oxygen
 321   Usage    : $pp->num_oxygen()
 322   Function : Retrieves the number of oxygen atoms in a sequence
 323   Returns  : Returns the number of oxygen atoms in a sequence
 324   Args     : none
 325
 326 =cut
 327
 328
 329 sub num_oxygen{
 330         my $self=shift;
 331
 332         ($self->{'oxy'}) = $self->{'output'}=~/Oxygen\s+O\s+(\d+)/;
 333
 334         return $self->{'oxy'};
 335
 336 }
 337
 338 =head2 num_sulphur
 339
 340   Title    : num_sulphur
 341   Usage    : $pp->num_sulphur()
 342   Function : Retrieves the number of sulphur atoms in a sequence
 343   Returns  : Returns the number of sulphur atoms in a sequence
 344   Args     : none
 345
 346 =cut
 347
 348
 349 sub num_sulphur{
 350         my $self=shift;
 351
 352         ($self->{'sul'}) = $self->{'output'}=~/Sulfur\s+S\s+(\d+)/;
 353
 354         return $self->{'sul'};
 355 }
 356
 357 =head2 half_life
 358
 359   Title    : half_life
 360   Usage    : $pp->half_life()
 361   Function : Retrieves the half life of a sequence
 362   Returns  : Returns the half life of a sequence
 363   Args     : none
 364
 365 =cut
 366
 367
 368 sub half_life{
 369         my $self=shift;
 370
 371         ($self->{'half_life'}) = $self->{'output'}=~/The estimated half-life is.*?(-{0,1}\d*\.{0,1}\d*)\s*hours \(mammalian reticulocytes, in vitro\)/;
 372
 373         return $self->{'half_life'};
 374 }
 375
 376 =head2 instability_index
 377
 378   Title    : instability_index
 379   Usage    : $pp->instability_index()
 380   Function : Retrieves the instability index of a sequence
 381   Returns  : Returns the instability index of a sequence
 382   Args     : none
 383
 384 =cut
 385
 386
 387 sub instability_index{
 388         my $self=shift;
 389
 390         ($self->{'InstabilityIndex'})=$self->{'output'}=~/The instability index \(II\) is computed to be (-{0,1}\d*\.{0,1}\d*)/;
 391         return $self->{'InstabilityIndex'};
 392 }
 393
 394 =head2 stability
 395
 396   Title    : stability
 397   Usage    : $pp->stability()
 398   Function : Calculates whether the sequence is stable or unstable
 399   Returns  : 'stable' or 'unstable'
 400   Args     : none
 401
 402 =cut
 403
 404
 405 sub stability{
 406         my $self=shift;
 407
 408         ($self->{'Stability'})=$self->{'output'}=~/This classifies the protein as\s(\w+)\./;
 409
 410         return $self->{'Stability'};
 411
 412 }
 413
 414 =head2 aliphatic_index
 415
 416   Title    : aliphatic_index
 417   Usage    : $pp->aliphatic_index()
 418   Function : Retrieves the aliphatic index of the sequence
 419   Returns  : Returns the aliphatic index of the sequence
 420   Args     : none
 421
 422 =cut
 423
 424
 425 sub aliphatic_index{
 426
 427         my $self=shift;
 428         ($self->{'AliphaticIndex'})=$self->{'output'}=~/<B>Aliphatic index:<\/B>\s*(-{0,1}\d*\.{0,1}\d*)/;
 429         return $self->{'AliphaticIndex'};
 430
 431 }
 432
 433 =head2 gravy
 434
 435   Title    : gravy
 436   Usage    : $pp->gravy()
 437   Function : Retrieves the grand average of hydropathicity (GRAVY) of a sequence
 438   Returns  : Returns the grand average of hydropathicity (GRAVY) of a sequence
 439   Args     : none
 440
 441 =cut
 442
 443
 444 sub gravy{
 445         my $self=shift;
 446
 447         ($self->{'GRAVY'})=$self->{'output'}=~/<B>Grand average of hydropathicity \(GRAVY\):<\/B>\s*(-{0,1}\d*\.{0,1}\d*)/;
 448         return $self->{'GRAVY'};
 449 }
 450
 451 =head2 AA_comp
 452
 453   Title    : AA_comp
 454   Usage    : $pp->AA_comp('P')
 455   Function : Retrieves the percentage composition of a given amino acid for a sequence
 456   Returns  : Returns the percentage composition of a given amino acid for a sequence
 457   Args     : A single letter amino acid code eg A, R, G, P etc
 458
 459 =cut
 460
 461
 462 sub AA_comp{
 463
 464         my $self=shift;
 465         my $aa=shift;
 466
 467         $aa=uc($aa);
 468
 469         my $AA={qw(A Ala R Arg N Asn D Asp C Cys Q Gln E Glu G Gly H His I Ile L Leu K Lys M Met F Phe P Pro S Ser T Thr W Trp Y Tyr V Val B Asx Z Glx X Xaa)};
 470
 471         ($self->{$aa})= $self->{'output'}=~/$AA->{$aa} \($aa\)\s+\d+\s+(\d+\.\d+)%/;
 472
 473         return $self->{$aa};
 474 }
 475
 476
 477 1;
 478
 479
 480
 481
 482
 483
 484
 485
 486