lib/CXGN/Tools/Pubmed.pm

   1 package CXGN::Tools::Pubmed;
   2 use strict;
   3 use warnings;
   4 use XML::Twig;
   5 use CXGN::Chado::Publication;
   6 use CXGN::Chado::Pubauthor;
   7
   8 =head1 CXGN::Tools::Pubmed
   9
  10 get data from pubmed and parse the fields that should be loaded in Chado schema
  11
  12
  13 =head2
  14
  15
  16 =head1 Author
  17
  18 Naama Menda
  19
  20 =cut
  21
  22 =head2 new
  23
  24  Usage: my $pubmed = CXGN::Tools::Pubmed->new($publication_obj);
  25  Desc:
  26  Ret:
  27  Args: $publication_object
  28  Side Effects:
  29  Example:
  30
  31 =cut
  32
  33 our $pub_object=undef;
  34
  35 sub new {
  36     my $class = shift;
  37     $pub_object= shift;
  38
  39     my $args = {};
  40     my $self = bless $args, $class;
  41
  42
  43     $self->set_pub_object($pub_object);
  44
  45     my $accession= $pub_object->get_accession();
  46     if ($accession) {
  47         $self->fetch($accession);
  48     }
  49     return $self;
  50 }
  51
  52
  53 sub fetch {
  54     my $self=shift;
  55     my $accession=shift;
  56     my $pub_xml = `wget "eutils.ncbi.nlm.nih.gov/entrez/eutils/efetch.fcgi?db=pubmed&id=$accession&rettype=xml&retmode=text"  -O -  `;
  57
  58     eval {
  59         my $twig=XML::Twig->new(
  60                                 twig_roots   =>
  61                                 {
  62                                     'Article/ArticleTitle'    => \&title,
  63                                     'JournalIssue/Volume'     => \&volume,
  64                                     'JournalIssue/Issue'      => \&issue,
  65                                     'DateCompleted/Year'      => \&pyear,
  66                                     'PubDate/Year'            => \&pyear,
  67                                     'Pagination/MedlinePgn'   => \&pages,
  68                                     'Journal/Title'           => \&journal_name,
  69                                     'PublicationTypeList/PublicationType'  => \&pub_type,
  70                                     'Abstract/AbstractText'   => \&abstract,
  71                                      Author                   => \&author,
  72                                      ELocationID     => \&e_id,
  73                                 },
  74                                 twig_handlers =>
  75                                 {
  76                                     # AbstractText     => \&abstract,
  77                                 },
  78
  79                                 pretty_print => 'indented',  # output will be nicely formatted
  80                                 );
  81
  82         $twig->parse($pub_xml ); # build it
  83
  84
  85         my $uniquename= $accession . ":" . $self->get_pub_object->get_title();
  86         if (!$pub_object->get_uniquename() ) { $pub_object->set_uniquename($uniquename); }
  87         #$pub_object->set_db_name('PMID');
  88         if (!$pub_object->get_cvterm_name) {
  89             $pub_object->set_cvterm_name('journal');
  90             $pub_object->d("!No cvtermname found for publication $accession. Setting to default 'journal'! ");
  91         }
  92     };
  93      if($@) {
  94         my $message= "Error in transaction or NCBI server seems to be down. Please check your input for accession $accession or try again later.\n $@";
  95         $pub_object->set_message($message);
  96     }else { return undef ; }
  97 }
  98
  99 =head2 get_pub_object
 100
 101  Usage:
 102  Desc:
 103  Ret:
 104  Args:
 105  Side Effects:
 106  Example:
 107
 108 =cut
 109
 110 sub get_pub_object {
 111   my $self=shift;
 112   return $self->{pub_object};
 113
 114 }
 115
 116 =head2 set_pub_object
 117
 118  Usage:
 119  Desc:
 120  Ret:
 121  Args:
 122  Side Effects:
 123  Example:
 124
 125 =cut
 126
 127 sub set_pub_object {
 128   my $self=shift;
 129   $self->{pub_object}=shift;
 130 }
 131
 132 =head2 get_title
 133
 134  Usage:
 135  Desc: get the title of the article
 136  Ret:
 137  Args:
 138  Side Effects:
 139  Example:
 140
 141 =cut
 142
 143 sub get_title {
 144     my $self=shift;
 145     return $self->{articleTitle};
 146 }
 147
 148 =head2 title
 149
 150  Usage:
 151  Desc: set the title of the article
 152  Ret:
 153  Args:
 154  Side Effects:
 155  Example:
 156
 157 =cut
 158
 159 sub title {
 160
 161      my ($twig, $elt)= @_;
 162      $pub_object->set_title($elt->text) ;
 163      $twig->purge;
 164  }
 165
 166
 167 =head2 volume
 168
 169  Usage:
 170  Desc:
 171  Ret:
 172  Args:
 173  Side Effects:
 174  Example:
 175
 176 =cut
 177
 178 sub volume {
 179     my ($twig, $elt)= @_;
 180     $pub_object->set_volume($elt->text) ;
 181     $twig->purge;
 182 }
 183
 184
 185 =head2 get_issue
 186
 187  Usage:
 188  Desc:
 189  Ret:
 190  Args:
 191  Side Effects:
 192  Example:
 193
 194 =cut
 195
 196 sub get_issue {
 197   my $self=shift;
 198   return $self->{issue};
 199
 200 }
 201
 202 =head2 issue
 203
 204  Usage:
 205  Desc:
 206  Ret:
 207  Args:
 208  Side Effects:
 209  Example:
 210
 211 =cut
 212
 213 sub issue {
 214     my ($twig, $elt)= @_;
 215     $pub_object->set_issue($elt->text) ;
 216
 217     $twig->purge;
 218 }
 219
 220 =head2 get_pyear
 221
 222  Usage:
 223  Desc:
 224  Ret:
 225  Args:
 226  Side Effects:
 227  Example:
 228
 229 =cut
 230
 231 sub get_pyear {
 232   my $self=shift;
 233   return $self->{pyear};
 234
 235 }
 236
 237 =head2 pyear
 238
 239  Usage:
 240  Desc:
 241  Ret:
 242  Args:
 243  Side Effects:
 244  Example:
 245
 246 =cut
 247
 248 sub pyear {
 249     my ($twig, $elt)= @_;
 250     my $pyear = $elt->text;
 251     $pub_object->set_pyear($pyear) if !$pub_object->get_pyear() ;
 252     #$pub_object->d("DEBUG:pyear is '$pyear'!!!!");
 253     $twig->purge;
 254 }
 255
 256 =head2 get_pages
 257
 258  Usage:
 259  Desc:
 260  Ret:
 261  Args:
 262  Side Effects:
 263  Example:
 264
 265 =cut
 266
 267 sub get_pages {
 268   my $self=shift;
 269   return $self->{pages};
 270
 271 }
 272
 273 =head2 pages
 274
 275  Usage:
 276  Desc:
 277  Ret:
 278  Args:
 279  Side Effects:
 280  Example:
 281
 282 =cut
 283
 284 sub pages {
 285     my ($twig, $elt)= @_;
 286     $pub_object->set_pages($elt->text) ;
 287
 288     $twig->purge;
 289 }
 290
 291
 292
 293
 294 =head2 get_journal_name
 295
 296  Usage:
 297  Desc:
 298  Ret:
 299  Args:
 300  Side Effects:
 301  Example:
 302
 303 =cut
 304
 305 sub get_journal_name {
 306   my $self=shift;
 307   return $self->{journal_name};
 308
 309 }
 310
 311 =head2 journal_name
 312
 313  Usage:
 314  Desc:
 315  Ret:
 316  Args:
 317  Side Effects:
 318  Example:
 319
 320 =cut
 321
 322 sub journal_name {
 323     my ($twig, $elt)= @_;
 324     $pub_object->set_series_name($elt->text) ;
 325
 326     $twig->purge;
 327 }
 328
 329 =head2 get_pub_type
 330
 331  Usage:
 332  Desc:
 333  Ret:
 334  Args:
 335  Side Effects:
 336  Example:
 337
 338 =cut
 339
 340 sub get_pub_type {
 341   my $self=shift;
 342   return $self->{pub_type};
 343
 344 }
 345
 346
 347 =head2 pub_type
 348
 349  Usage:
 350  Desc:
 351  Ret:
 352  Args:
 353  Side Effects:
 354  Example:
 355
 356 =cut
 357
 358 sub pub_type {
 359     my ($twig, $elt)= @_;
 360     my $pub_type= $elt->text;
 361     if ($pub_type =~ m/Journal/i) {
 362         $pub_type = 'journal' ;
 363         $pub_object->set_cvterm_name($pub_type) ;
 364         $pub_object->d("DEBUG:found pub_type '$pub_type' . Setting cvterm_name to 'journal' !!\n");
 365
 366     }else {
 367         #$pub_object->d("DEBUG:pub_type $pub_type is not a recognizeable cvterm!!\n");
 368     }
 369     $twig->purge;
 370 }
 371
 372
 373 =head2 get_abstract
 374
 375  Usage:
 376  Desc:
 377  Ret:
 378  Args:
 379  Side Effects:
 380  Example:
 381
 382 =cut
 383
 384 sub get_abstract {
 385   my $self=shift;
 386   return $self->{abstract};
 387
 388 }
 389
 390 =head2 abstract
 391
 392  Usage:
 393  Desc:
 394  Ret:
 395  Args:
 396  Side Effects:
 397  Example:
 398
 399 =cut
 400
 401 sub abstract {
 402     my ($twig, $elt)= @_;
 403     $pub_object->set_abstract($elt->text) ;
 404
 405     $twig->purge;
 406 }
 407
 408
 409
 410 =head2 author
 411
 412  Usage:
 413  Desc:
 414  Ret:
 415  Args:
 416  Side Effects:
 417  Example:
 418
 419 =cut
 420
 421 sub author {
 422     my ($twig, $elt)= @_;
 423
 424     my $lastname=$elt->children_text('LastName');
 425     my $initials=$elt->children_text('Initials');  #sometimes the firstname has no initials but full first name 'ForName'..
 426
 427     if (!$initials) {  $initials=$elt->children_text('ForeName') || $elt->children_text('FirstName') ; }
 428
 429
 430     my $author_data=  $lastname ."|" . $initials ; #.",".$initials ;
 431
 432     $pub_object->add_author($author_data) ;
 433     #$rank ++;
 434     $twig->purge;
 435 }
 436
 437 =head2 e_id
 438
 439  Usage:
 440  Desc: Store DOI
 441  Ret:
 442  Args:
 443  Side Effects: sets eid for the publication object
 444  Example:
 445
 446 =cut
 447
 448 sub e_id {
 449     my ($twig, $elt)= @_;
 450     $pub_object->set_eid($elt->text) ;
 451     $twig->purge;
 452 }
 453
 454
 455
 456 1;