cgi-bin/tools/blast/index.pl

   1 use CatalystX::GlobalContext qw( $c );
   2 =head1 NAME
   3
   4 /tools/blast/ - the entry page to the SGN blast tool
   5
   6 =head1 DESCRIPTION
   7
   8 This page displays a text box for a query sequence, along with a
   9 number of pull down menus to select (1) the dataset to match against,
  10 (2) the blast program, (3) an evalue cutoff, (4) a substitution
  11 matrix, (5) the number of maximum hits to show, and (6) parameters
  12 affecting the output of the results.
  13
  14 Important features: the default dataset selected is
  15 tomato_combined. If the user selects another dataset, this is stored
  16 in the user prefs and comes up as the default later on. A specific
  17 dataset can be pre-selected in a link using the db_id parameter.
  18
  19 This script is in desparate need of a decent re-factoring...
  20
  21 Parameters:
  22
  23 =over 5
  24
  25 =item preload_id
  26
  27 an id that can be preloaded from the database. Requires preload_type
  28 to be set as well.
  29
  30 =item preload_type
  31
  32 the type of identifier to preload from the database (7 is EST, 15 is
  33 unigene). Requires preload_id to be set as well.
  34
  35 =item seq
  36
  37 a sequence to blast, automatically filled in
  38
  39 =item interface_type
  40
  41 either 'simple' or 'advanced', affects the features displayed on the
  42 page.
  43
  44 =item db_id
  45
  46 an id of a database, which will appear pre-selected in the pulldown
  47 menu.
  48
  49 =back
  50
  51 =head1 AUTHORS
  52
  53 This script was initially written by Koni Wright. Many additions and
  54 changes by SGN staff. POD by Lukas.
  55
  56 =cut
  57
  58
  59 use strict;
  60 use POSIX;
  61 use English;
  62
  63 use Memoize;
  64 use Storable qw/ retrieve nstore /;
  65
  66 use Tie::UrlEncoder;
  67 our %urlencode;
  68
  69 use File::Flock;
  70
  71 use CXGN::Page;
  72 use CXGN::BlastDB;
  73 use CXGN::Page::FormattingHelpers qw/page_title_html modesel info_table_html hierarchical_selectboxes_html simple_selectbox_html/;
  74 use CXGN::Page::UserPrefs;
  75 use CXGN::Tools::List qw/evens distinct/;
  76 use CatalystX::GlobalContext '$c';
  77
  78 my $page  = CXGN::Page->new("BLAST Search Interface","Evan");
  79 my $dbh   = CXGN::DB::Connection->new;
  80 my $prefs = CXGN::Page::UserPrefs->new( $dbh );
  81
  82 my %params;
  83 @params{qw/preload_id preload_type seq interface_type db_id flush_cache/} = $page->get_encoded_arguments("preload_id","preload_type","seq","interface_type", "db_id", 'flush_cache');
  84
  85 $params{interface_type} ||= 0;
  86
  87
  88 my $blast_path = $c->config->{'blast_path'};
  89 my $blast_version = do {
  90   unless( -x "$blast_path/blastall") {
  91     ''
  92   } else {
  93     eval {
  94       open BP, qq{echo '>shutup' | $blast_path/blastall -p blastn 2>&1 |};
  95       my $v = '';
  96       while (<BP>) {
  97         if (m/BLASTN (.+)/) {
  98           $v = " v. $1";
  99           last;
 100         }
 101       }
 102       close BP;
 103     }; if( $EVAL_ERROR ) {
 104       ''
 105     }
 106   }
 107 };
 108
 109 my $preload_seq;
 110 if ($params{preload_type}) {
 111   if ($params{preload_type} == 7) {
 112
 113     my $estq = $dbh->prepare_cached(<<EOSQL);
 114 SELECT COALESCE( CASE WHEN hqi_length > 0 THEN SUBSTRING(seq, hqi_start::integer+1, hqi_length::integer)
 115                       ELSE NULL
 116                  END,
 117                  seq
 118                )
 119 FROM est
 120 LEFT JOIN qc_report USING (est_id)
 121 WHERE est.est_id=?
 122 EOSQL
 123
 124     $estq->execute($params{preload_id});
 125
 126     if ($estq->rows == 0) {
 127       die("Preloaded BLAST search specified EST sequence SGN-E$params{preload_id} but it is not found in database ($params{preload_id})");
 128     }
 129
 130     ($preload_seq) = $estq->fetchrow_array();
 131
 132     my $x = "";
 133     my $i = 0;
 134     while((length($preload_seq) - $i) > 78) {
 135       $x .= substr($preload_seq, $i, 78) . "\n";
 136       $i+=78;
 137     }
 138     $x .= substr($preload_seq, $i) . "\n";
 139     $preload_seq = ">SGN-E$params{preload_id}\n$x\n";
 140
 141   } elsif ($params{preload_type} == 15) {
 142
 143     my $memberq = $dbh->prepare_cached("SELECT nr_members FROM unigene WHERE unigene_id=?");
 144     $memberq->execute($params{preload_id});
 145     if ($memberq->rows == 0) {
 146       $page->error_page("Preloaded BLAST search specified unigene identifier ($params{preload_id}) which is not found");
 147     }
 148
 149     my ($nr_members) = $memberq->fetchrow_array();
 150     if ($nr_members > 1) {
 151       ($preload_seq) = $dbh->selectrow_array(<<EOSQL,undef,$params{preload_id})
 152 SELECT seq
 153 FROM unigene
 154 INNER JOIN unigene_consensi
 155 USING (consensi_id)
 156 WHERE unigene_id=?
 157 EOSQL
 158     } else {
 159       ($preload_seq) = $dbh->selectrow_array(<<EOSQL,undef,$params{preload_id});
 160 SELECT COALESCE( CASE WHEN hqi_length > 0
 161                            THEN SUBSTRING(seq, hqi_start::integer+1, hqi_length::integer)
 162                       ELSE NULL
 163                  END,
 164                  seq
 165                )
 166 FROM unigene
 167 LEFT JOIN unigene_member USING (unigene_id)
 168 LEFT JOIN est USING (est_id)
 169 LEFT JOIN qc_report USING (est_id)
 170 WHERE unigene.unigene_id=?
 171 EOSQL
 172     }
 173     my $x = "";
 174     my $i = 0;
 175     while((length($preload_seq) - $i) > 78) {
 176       $x .= substr($preload_seq, $i, 78) . "\n";
 177       $i+=78;
 178     }
 179     $x .= substr($preload_seq, $i) . "\n";
 180     $preload_seq = ">SGN-U$params{preload_id}\n$x";
 181   } else {
 182     $page->error_page("Unknown preloaded sequence type\n");
 183   }
 184 } elsif ($params{seq}) {
 185   $preload_seq = $params{seq};
 186 } else {
 187   $preload_seq = "";
 188 }
 189
 190 $page->header('SGN BLAST');
 191 $page->jsan_use('jquery');
 192
 193 my ($databases,$programs,$programs_js) = blast_db_prog_selects($params{db_id});
 194 my $spellcheck_js = <<'';
 195 // turn off spell check on sequence inputs without emitting invalid HTML
 196 jQuery(function($) { $('#sequence_input').attr('spellcheck',false) });
 197
 198
 199 sub hash2param {
 200   my %args = @_;
 201   return join '&', map "$urlencode{$_}=$urlencode{$args{$_}}", distinct evens @_;
 202 }
 203
 204 print page_title_html("NCBI BLAST$blast_version");
 205 print modesel([ ['?'.hash2param(%params, interface_type => 0),'Simple'],
 206                 ['?'.hash2param(%params, interface_type => 1),'Advanced'],
 207               ],
 208               $params{interface_type},
 209              );
 210
 211 #simple blast interface form
 212 if($params{interface_type} == 0) {
 213   print <<EOF
 214
 215 <script language="JavaScript" type="text/JavaScript" >
 216
 217 function clearField() {
 218     // OK - there are three ways to clear the fields.
 219     // The first is let the browser do it. Then it does not clear
 220     // with preset sequences. The second way is to clear it with javascript.
 221     // on reload, this will create confusion with preset sequences.
 222     // The third way is simply to redirect to the empty page. Thats good
 223     // because it also resets to the users preferred datatset.
 224     // var i = document.getElementById("sequence_input");
 225     // i.innerHTML='';
 226     window.location="index.pl";
 227 }
 228
 229 </script>
 230
 231 <form method="post" action="blast_result.pl" name="blastform">
 232   <input type="hidden" checked="checked" name="filterq" value="1" />
 233   <input type="hidden" name="interface_type" value="simple" />
 234   <input type="hidden" name="outformat" value="0" />
 235   <table align="center" summary="" cellpadding="0" cellspacing="15">
 236     <tr><td><b>Sequence Set</b> </td><td>$databases <a style="font-size: 80%" title="View details of each database" href="dbinfo.pl">db details</a></td></tr>
 237     <tr><td><b>Program</b> </td><td>$programs</td></tr>
 238     <tr><td colspan="2" align="center"><b>Query sequence</b><div style="font-size: 80%">single sequence only, use Advanced for multiple</div><textarea name="sequence" id="sequence_input" rows="8" cols="80">$preload_seq</textarea></td></tr>
 239     <tr><td colspan="2">
 240       <table width="100%"><tr>
 241          <td>
 242 EOF
 243   .info_table_html('Expect (e-value) Threshold' => '<input type="text" size="6" value="1e-10" name="expect" />',
 244                    'Substitution Matrix' => simple_selectbox_html( name => 'matrix',
 245                                                                     choices => [ [ 'BLOSUM62', 'BLOSUM62 (default)' ],
 246                                                                                  [ 'BLOSUM80', 'BLOSUM80 (recent divergence)' ],
 247                                                                                  [ 'BLOSUM45', 'BLOSUM80 (ancient divergence)' ],
 248                                                                                  'PAM30',
 249                                                                                  'PAM70',
 250                                                                                ],
 251                                                                   ),
 252                     'Max. hits to show' => '<input type="text" name="maxhits" size="6" value="100" />',
 253                    ' ' => '<div style="text-align: right"><input type="reset" value="Clear" onclick="clearField(); "/> <input type="submit" name="search" value="Search" /></div>',
 254                     'Show Graphics' => simple_selectbox_html( name => 'output_graphs',
 255                                                               choices => [ [ 'bioperl_histogram', 'all' ],
 256                                                                            'none',
 257                                                                            [ 'bioperl_only', 'alignment summary only' ],
 258                                                                            [ 'histogram_only', 'conservedness histogram only' ],
 259                                                                          ],
 260                                                             ),
 261                     __multicol => 2,
 262                     __border => 0,
 263                     __tableattrs => 'width="100%"',
 264                   )
 265     .<<EOF
 266          </td>
 267       </tr>
 268       </table>
 269     </td></tr>
 270   </table>
 271 </form>
 272 <script language="JavaScript" type="text/javascript">
 273 $programs_js
 274 $spellcheck_js
 275 </script>
 276 EOF
 277       ;
 278 }
 279
 280 #advanced blast interface form
 281 else {
 282   my $mselect = simple_selectbox_html( name => 'outformat',
 283                                        choices => [ [0  => '0 - pairwise (default)'],
 284                                                     [1  => '1 - query-anchored showing identities'],
 285                                                     [2  => '2 - query-anchored no identities'],
 286                                                     [3  => '3 - flat query-anchored, show identities'],
 287                                                     [4  => '4 - flat query-anchored, no identities'],
 288                                                     [5  => '5 - query-anchored no identities and blunt ends'],
 289                                                     [6  => '6 - flat query-anchored, no identities and blunt ends'],
 290                                                     [7  => '7 - XML Blast output'],
 291                                                     [8  => '8 - tabular'],
 292                                                     [9  => '9 - tabular with comment lines'],
 293                                                     [10 => '10 - ASN, text'],
 294                                                     [11 => '11 - ASN, binary'],
 295                                                   ],
 296                                      );
 297
 298   print <<EOF;
 299 <table style="border: 1px solid gray; padding: 1em 2em 1em 2em; background: #eeeeff;"><tr><td><img src="/documents/img/info_icon.png" border="0" style="margin-right: 1em; vertical-align: middle" /></td><td style="padding-top: 0.35em">This version of the BLAST online tool allows multiple query sequences, more control over running options, and more report formats.</td></tr></table>
 300 <form method="post" action="blast_result.pl" name="blastform" enctype="multipart/form-data">
 301   <input type="hidden" name="interface_type" value="advanced" />
 302   <table id="blastinput" align="center" summary="" cellpadding="0" cellspacing="15">
 303     <tr><td><b>Database (<tt>-d</tt>)</b> </td><td>$databases <a style="font-size: 80%" title="View details of each database" href="dbinfo.pl">db details</a></td></tr>
 304     <tr><td><b>Program (<tt>-p</tt>)</b> </td><td>$programs</td></tr>
 305     <tr>
 306       <td><b>Query sequences (<tt>-i</tt>)</b></td>
 307       <td >
 308         <textarea class="fix" id="sequence_input" name="sequence" rows="8" cols="65">$preload_seq</textarea><br />
 309         <b>AND/OR upload multi-fasta query file</b> <input type="file" name="file" />
 310       </td>
 311     </tr>
 312     <tr><td><b>Output format (<tt>-m</tt>)</b></td>
 313         <td>$mselect</td>
 314     </tr>
 315     <tr><td><b>Substitution Matrix (<tt>-M</tt>)</b></td>
 316         <td >
 317             <select name="matrix">
 318             <option value="BLOSUM62">BLOSUM62 (default)</option>
 319             <option value="BLOSUM80">BLOSUM80 (recent divergence)</option>
 320             <option value="BLOSUM45">BLOSUM45 (ancient divergence)</option>
 321             <option value="PAM30">PAM30</option>
 322             <option value="PAM70">PAM70</option>
 323             </select>
 324         </td>
 325     </tr>
 326     <tr><td><b>Expectation value (<tt>-e</tt>)</b> </td>
 327         <td ><input type="text" size="6" value="1e-10" name="expect" /></td>
 328     </tr>
 329     <tr><td><b>Max DB seqs to show hits from (<tt>-b</tt>)</b></td>
 330         <td ><input type="text" name="maxhits" size="6" value="100" /></td>
 331     </tr>
 332     <tr><td><b>Filter query sequence (DUST with blastn, SEG with others) (<tt>-F</tt>)</b></td>
 333         <td><input type="checkbox" checked="checked" name="filterq" /></td>
 334     </tr>
 335     <tr>
 336         <td>
 337                 <b>Show Graphics</b><br /><span style="font-size: 80%">not available for multiple query seqs</span>
 338         </td>
 339         <td>
 340                 <select name="output_graphs">
 341                         <option value="none">none</option>
 342                         <option value="bioperl_only">alignment summary only</option>
 343                         <option value="histogram_only">conservedness histogram only</option>
 344                         <option value="bioperl_histogram" selected="selected">all</option>
 345                 </select>
 346         </td>
 347     </tr>
 348
 349     <tr><td align="right"><input type="reset" value="Clear" /></td><td align="center"><input type="submit" name="search" value="Submit" style="background: red; font-size: 130%" /></td></tr>
 350      </table>
 351 </form>
 352 <script language="JavaScript" type="text/javascript">
 353 $programs_js
 354 $spellcheck_js
 355 </script>
 356 EOF
 357
 358 }
 359
 360 $page->footer();
 361
 362 ##########################################################################################################################
 363
 364 memoize '_cached_file_modtime';
 365 sub _cached_file_modtime {
 366   shift->file_modtime
 367 }
 368
 369 sub blast_db_prog_selects {
 370     my $db_id = shift;
 371
 372     my $db_choices = blast_db_choices();
 373
 374     return '<span class="ghosted">The BLAST service is temporarily unavailable, we apologize for the inconvenience</span>'
 375         unless @$db_choices;
 376
 377     # DB select box will either the db_id supplied, or what the user last selected, or the tomato combined blast db
 378     my $selected_db_id = $db_id #|| $prefs->get_pref('last_blast_db_id')
 379         || do {
 380             my ($d) = map $_->blast_db_id,
 381                       grep _cached_file_modtime($_),
 382                       grep $_->web_interface_visible,
 383                       CXGN::BlastDB->search_ilike( title => '%SGN Tomato Combined%' );
 384             $d;
 385         };
 386
 387     my %prog_descs = ( blastn  => 'BLASTN (nucleotide to nucleotide)',
 388                        blastx  => 'BLASTX (nucleotide to protein; query translated to protein)',
 389                        blastp  => 'BLASTP (protein to protein)',
 390                        tblastx => 'TBLASTX (protein to protein; both database and query are translated)',
 391                        tblastn => 'TBLASTN (protein to nucleotide; database translated to protein)',
 392                       );
 393
 394     my @program_choices = map {
 395         my ($db) = @$_;
 396         if ($db->type eq 'protein') {
 397             [map [$_,$prog_descs{$_}], 'blastx','blastp']
 398         } else {
 399             [map [$_,$prog_descs{$_}], 'blastn','tblastx','tblastn']
 400         }
 401     } grep ref, @$db_choices;
 402
 403     @$db_choices = map {ref($_) ? $_->[1] : $_} @$db_choices;
 404
 405     return hierarchical_selectboxes_html( parentsel => { name => 'database',
 406                                                          choices => $db_choices,
 407                                                          ( $selected_db_id ? (selected => $selected_db_id) : () ),
 408                                                        },
 409                                           childsel  => { name => 'program' },
 410                                           childchoices => \@program_choices
 411                                          );
 412 }
 413
 414 sub blast_db_choices {
 415
 416     my $choices_cache_filename = $c->path_to( $c->generated_file_uri('blast','choices_cache.dat') );
 417     my $lockfile = "$choices_cache_filename.lock";
 418
 419     unless( $params{flush_cache} ) {
 420         my $l = File::Flock->new($lockfile,'shared');
 421         my $cache_modtime = (stat($choices_cache_filename))[9];
 422
 423         if( $cache_modtime && $cache_modtime > time - 15*60 ) {
 424             my $data = retrieve( $choices_cache_filename );
 425             return $data if $data;
 426         }
 427     }
 428
 429     my $l = File::Flock->new($lockfile);
 430     my $choices = _build_blast_db_choices();
 431     nstore( $choices, $choices_cache_filename )
 432         or warn "WARNING: $! caching blast db choices in file '$choices_cache_filename'";
 433     return $choices;
 434 }
 435
 436 sub _build_blast_db_choices {
 437
 438     sleep 5;
 439
 440     my @db_choices = map {
 441         my @dbs = map [ $_, bdb_opt($_) ],
 442             grep _cached_file_modtime($_), #filter for dbs that are on disk
 443                 $_->blast_dbs( web_interface_visible => 't'); #get all dbs in this group
 444         @dbs ? ('__'.$_->name, @dbs) : ()
 445     } CXGN::BlastDB::Group->search_like(name => '%',{order_by => 'ordinal, name'});
 446
 447     my @ungrouped_dbs =
 448         grep _cached_file_modtime($_),
 449         CXGN::BlastDB->search(
 450             blast_db_group_id => undef,
 451             web_interface_visible => 't',
 452             { order_by => 'title' }
 453            );
 454
 455     push @db_choices, ( '__Other',
 456                         map [$_,bdb_opt($_)],
 457                         @ungrouped_dbs
 458                        );
 459
 460     return \@db_choices;
 461 }
 462
 463 sub bdb_opt {
 464     my $db = shift;
 465 #     my $timestamp = _cached_file_modtime($db)
 466 #         or return;
 467 #     $timestamp = strftime(' &nbsp;(%F)',gmtime _cached_file_modtime($db));
 468     my $seq_count = $db->sequences_count;
 469
 470     [$db->blast_db_id, $db->title ]
 471 }