lib/CXGN/Cview/MapFactory/SGN.pm

   1
   2 =head1 NAME
   3
   4 CXGN::Cview::MapFactory - a factory object for CXGN::Cview::Map objects
   5
   6 =head1 SYNOPSYS
   7
   8 my $map_factory  = CXGN::Cview::MapFactory->new($dbh);
   9 $map = $map_factory->create({map_version_id=>"u1"});
  10
  11 =head1 DESCRIPTION
  12
  13 see L<CXGN::Cview::MapFactory>.
  14
  15 The MapFactory constructor takes a database handle (preferably constructed using CXGN::DB::Connection object). The map objects can then be constructed using the create function, which takes a hashref as a parameter, containing either map_id or map_version_id as a key (but not both). map_ids will be converted to map_version_ids immediately. Map_version_ids are then analyzed and depending on its format, CXGN::Cview::Map object of the proper type is returned.
  16
  17 The function get_all_maps returns all maps as list of appropriate CXGN::Cview::Map::* objects.
  18
  19 For the current SGN implementation, the following identifier formats are defined and yield following corresponding map objects
  20
  21  \d+       refers to a map id in the database and yields either a
  22            CXGN::Cview::Map::SGN::Genetic (type genetic)
  23            CXGN::Cview::Map::SGN::FISH (type fish)
  24            CXGN::Cview::Map::SGN::Sequence (type sequence)
  25  u\d+      refers to a user defined map and returns:
  26            CXGN::Cview::Map::SGN::User object
  27  filepath  refers to a map defined in a file and returns a
  28            CXGN::Cview::Map::SGN::File object
  29  il\d+      refers to a population id in the phenome.population table
  30            (which must be of type IL) and returns a
  31            CXGN::Cview::Map::SGN::IL object
  32  p\d+      CXGN::Cview::Map::SGN::Physical
  33  c\d+      CXGN::Cview::Map::SGN::Contig
  34  o         CXGN::Cview::Map::SGN::ProjectStats map object
  35
  36 The actual map objects returned are defined in the CXGN::Cview::Maps namespace. Because this is really a compatibility layer, an additional namespace of the resource is appended, such that a genetic map at SGN could be defined as CXGN::Cview::Maps::SGN::Genetic . If no corresponding map is found, undef is returned.
  37
  38 =head1 AUTHOR(S)
  39
  40 Lukas Mueller (lam87@cornell.edu)
  41
  42 =head1 VERSION
  43
  44 1.0, March 2007
  45
  46 =head1 LICENSE
  47
  48 Refer to the L<CXGN::LICENSE> file.
  49
  50 =head1 FUNCTIONS
  51
  52 This class implements the following functions:
  53
  54 =cut
  55
  56 use strict;
  57
  58 package CXGN::Cview::MapFactory::SGN;
  59
  60 use base qw| CXGN::DB::Object |;
  61
  62 use Scalar::Util qw/blessed/;
  63
  64 use CXGN::Cview::Map::SGN::Genetic;
  65 #use CXGN::Cview::Map::SGN::User;
  66 use CXGN::Cview::Map::SGN::Fish;
  67 use CXGN::Cview::Map::SGN::Sequence;
  68 use CXGN::Cview::Map::SGN::IL;
  69 use CXGN::Cview::Map::SGN::Physical;
  70 use CXGN::Cview::Map::SGN::ProjectStats;
  71 use CXGN::Cview::Map::SGN::AGP;
  72 #use CXGN::Cview::Map::SGN::ITAG;
  73 use CXGN::Cview::Map::SGN::Contig;
  74 use CXGN::Cview::Map::SGN::Scaffold;
  75 use CXGN::Cview::Map::SGN::Image;
  76 use CXGN::Cview::Map::SGN::QTL;
  77
  78 =head2 function new()
  79
  80   Synopsis:     constructor
  81   Arguments:    a database handle
  82   Returns:      a CXGN::Cview::MapFactory::SGN object
  83   Side effects: none
  84   Description:  none
  85
  86 =cut
  87
  88 sub new {
  89     my $class = shift;
  90     my $dbh = shift;
  91     my $context = shift;
  92
  93     unless( blessed($context) && $context->isa('SGN::Context') ) {
  94         require SGN::Context;
  95         $context = SGN::Context->new();
  96     }
  97     my $self = $class->SUPER::new($dbh);
  98
  99     $self->{context}=$context;
 100
 101
 102     return $self;
 103 }
 104
 105 =head2 function create()
 106
 107   Description:  creates a map based on the hashref given, which
 108                 should either contain the key map_id or map_version_id
 109                 and an appropriate identifier. The function returns undef
 110                 if a map of the given id cannot be found/created.
 111   Example:
 112
 113 =cut
 114
 115 sub create {
 116     my $self = shift;
 117     my $hashref = shift;
 118     #print STDERR "Hashref = map_id => $hashref->{map_id}, map_version_id => $hashref->{map_version_id}\n";
 119
 120     my $c = $self->{context};
 121     my $temp_dir = $c->path_to( $c->config->{tempfiles_subdir} );
 122
 123     if (!exists($hashref->{map_id}) && !exists($hashref->{map_version_id})) {
 124         die "[CXGN::Cview::MapFactory] Need either a map_id or map_version_id.\n";
 125     }
 126     if ($hashref->{map_id} && $hashref->{map_version_id}) {
 127         die "[CXGN::Cview::MapFactory] Need either a map_id or map_version_id - not both.\n";
 128     }
 129     if ($hashref->{map_id}) {
 130         $hashref->{map_version_id}=CXGN::Cview::Map::Tools::find_current_version($self->get_dbh(), $hashref->{map_id});
 131     }
 132
 133     # now, we only deal with map_versions...
 134     #
 135     my $id = $hashref->{map_version_id};
 136
 137     #print STDERR "MapFactory: dealing with id = $id\n";
 138
 139     # if the map_version_id is purely numeric,
 140     # check if the map is in the maps table and generate the
 141     # appropriate map
 142
 143     if ($id=~/^\d+$/) {
 144         my $query = "SELECT map_version_id, map_type, map_id, short_name FROM sgn.map join sgn.map_version using(map_id) WHERE map_version_id=?";
 145         my $sth = $self->get_dbh()->prepare($query);
 146         $sth->execute($id);
 147         my ($id, $map_type) = $sth->fetchrow_array();
 148     $map_type ||= '';
 149         if ($map_type =~ /genetic/i) {
 150             return CXGN::Cview::Map::SGN::Genetic->new($self->get_dbh(), $id);
 151
 152         }
 153         elsif ($map_type =~ /fish/) {
 154             #print STDERR "Creating a fish map...\n";
 155             return CXGN::Cview::Map::SGN::Fish->new($self->get_dbh(), $id, { pachytene_file => $c->config->{'basepath'}."/documents/cview/pachytene/pachytene_stack.txt", });
 156         }
 157         elsif ($map_type =~ /seq/) {
 158             #print STDERR "Creating a seq map...\n";
 159             return CXGN::Cview::Map::SGN::Sequence->new($self->get_dbh(), $id);
 160         }
 161         elsif ($map_type =~ /qtl/i) {
 162
 163             my $qtl = CXGN::Cview::Map::SGN::QTL->new($self->get_dbh(), $id);
 164
 165             return $qtl;
 166         }
 167     }
 168     elsif ($id =~ /^u/i) {
 169         #return CXGN::Cview::Map::SGN::User->new($self->get_dbh(), $id);
 170     }
 171     elsif ($id =~ /^il/i) {
 172     my $abstract =
 173         "The tomato Introgression lines (ILs) are a set of nearly isogenic lines (NILs) developed by Dani Zamir through a succession of backcrosses, where each line carries a single genetically defined chromosome segment from a divergent genome. The ILs, representing whole-genome coverage of S. pennellii in overlapping segments in the genetic background of S. lycopersicum cv. M82, were first phenotyped in 1993, and presently this library consists of 76 genotypes. ";
 174
 175         my ($population_id, $map_id) = $self->get_db_ids($id);
 176         if ($map_id == 9) {
 177             $abstract .= " This IL map is based on markers of the F2-2000 map. ILs have also been mapped <a href=\"map.pl?map_id=il6.5&amp;show_ruler=1&amp;show_offsets=1\" >with the ExPEN1992 map as a reference</a>.";
 178         }
 179         elsif ($map_id ==5) {
 180             $abstract .= " The IL lines on this map have been characterized based on the markers on the 1992 tomato map. ILs have also been mapped <a href=\"map.pl?map_id=il6.9&amp;show_ruler=1&amp;show_offsets=1\" >with the ExPEN2000 map as a reference</a>. ";
 181         }
 182
 183         my $ref_map = "ExPEN2000";
 184         if ($id==5) { $ref_map = "ExPEN1992";}
 185         my $long_name =qq | Solanum lycopersicum Zamir Introgression Lines (IL) based on $ref_map |;
 186
 187         return CXGN::Cview::Map::SGN::IL->new($self->get_dbh(), $id,
 188                                               { short_name    => "Tomato IL map",
 189                                                 long_name     => $long_name,
 190                                                 abstract      => $abstract,
 191
 192
 193                                               });
 194     }
 195     elsif ($id =~ /^\//) {
 196         #return CXGN::Cview::Map::SGN::File->new($dbh, $id);
 197     }
 198     elsif ($id =~ /^p\d+/) {
 199         return CXGN::Cview::Map::SGN::Physical->new($self->get_dbh(), $id);
 200     }
 201     elsif ($id =~ /^o$/i) {
 202
 203         return CXGN::Cview::Map::SGN::ProjectStats->new($self->get_dbh(), {
 204             short_name=>"Tomato Sequencing Progress",
 205             long_name=>"Tomato Sequencing Statistics by Chromosome",
 206             abstract => $self->get_abstract(),
 207
 208         });
 209     }
 210
 211     elsif ($id =~ /^agp$/i) {
 212         return CXGN::Cview::Map::SGN::AGP->new($self->get_dbh(), $id, {
 213                                                short_name => "Tomato AGP map",
 214                                                long_name => "Tomato (Solanum lycopersicum) Accessioned Golden Path map",
 215                                                abstract => "<p>The AGP map shows the sequencing progress of the international tomato genome sequencing project by listing all finished clones by estimated physical map position . Each sequencing center generates one or more AGP (Accessioned Golden Path) files and uploads them to SGN. These files contain all the sequenced BACs, their position on the chromosome, the overlaps with other BACs and other information. For a complete specification, please refer to the <a href=\"http://www.sanger.ac.uk/Projects/C_elegans/DOCS/agp_files.shtml\">Sanger AGP specification</a>. The AGP files can also be downloaded from the SGN FTP site, at <a href=\"ftp://ftp.sgn.cornell.edu/tomato_genome/agp/\">ftp://ftp.sgn.cornell.edu/tomato_genome/agp/</a>.</p> <p>Note that this map is in testing (beta), and not all features may be functional.</p>" ,
 216                                                temp_dir => $temp_dir ,
 217                                                basedir       => $c->config->{"basepath"},
 218                                                documents_subdir => $c->config->{"tempfiles_subdir"},
 219                                                file => $c->config->{"static_content_path"}."/cview/agp/scaffolds/S_lycopersicum_chromosomes_from_scaffolds.2.40.agp",
 220                                                cache_dir => $c->config->{"basepath"}."/".$c->config->{tempfiles_subdir}."/cview/cache_file/",
 221
 222                                                },
 223             );
 224
 225     }
 226 #     elsif ($id =~ /^itag$/i) {
 227
 228 #       my (@sources) = map $_->data_sources(), $c->enabled_feature('gbrowse2');
 229 #       my ($gbrowse_itag) = grep $_->description()=~/ITAG_devel.+genomic/i, @sources;
 230 #       my @dbs;
 231 #       if ($gbrowse_itag) {
 232 #           @dbs = $gbrowse_itag->databases();
 233 #           @dbs > 1 and die "I can handle only one db!";
 234 #       }
 235
 236 #         return unless $gbrowse_itag;
 237
 238 #         my $gbrowse_view_link = $gbrowse_itag->view_url;
 239
 240 #       my $marker_link =  sub { my $id = shift; return "$gbrowse_view_link?name=$id"; };
 241 #       return CXGN::Cview::Map::SGN::ITAG->new($self->get_dbh(), $id, {
 242 #                                               short_name => "Tomato ITAG map",
 243 #                                               long_name=>"Tomato (Solanum lycopersicum) ITAG map",
 244 #                                               abstract=>"<p>The ITAG map shows the contig assembly and the corresponding BACs as used by the most recent annotation from the International Tomato Annotation Group (ITAG, see <a href=\"http://www.ab.wur.nl/TomatoWiki\">ITAG Wiki</a>). Clicking on the contigs will show the ITAG annotation in the genome browser.",
 245 #                                               temp_dir    => $temp_dir,
 246 #                                               marker_link => $marker_link,
 247
 248 #                                           }
 249 #           );
 250 #     }
 251
 252 #     elsif ($id =~ /scaffold103/) {
 253
 254 #       return CXGN::Cview::Map::SGN::Scaffold->new($self->get_dbh(), $id, {
 255 #           file=> '/data/prod/public/tomato_genome/wgs/chromosomes/assembly_1.03/chromosome_defs_v1.03_sorted.txt',
 256 #           abstract=>'test abstract',
 257 #           temp_dir=>$temp_dir,
 258 #           short_name=>'Tomato scaffold map V1.03',
 259 #           long_name=>'Solanum lycopersicum scaffold map V1.03',
 260 #           marker_link => sub {},
 261
 262 #                                                   } );
 263 #     }
 264
 265 #     elsif ($id =~ /scaffold100/) {
 266 #       my (@sources) = map $_->data_sources(), $c->enabled_feature('gbrowse2');
 267 #       my ($gbrowse) = grep $_->description()=~/ITAG1.+genomic/i, @sources;
 268 #       if (!$gbrowse) { die "No such map in GBrowse."; }
 269 #       my @dbs;
 270 #       if ($gbrowse) {
 271 #           @dbs = $gbrowse->databases();
 272 #           @dbs > 1 and die "I can handle only one db!";
 273 #       }
 274
 275 #         return unless $gbrowse;
 276
 277 #       my $gbrowse_view_link = $gbrowse->view_url;
 278
 279 #       my $marker_link =  sub { my $id = shift; return "$gbrowse_view_link?name=$id"; };
 280
 281 #       return CXGN::Cview::Map::SGN::Scaffold->new($self->get_dbh(), $id, {
 282 #           file=> '/data/prod/public/tomato_genome/wgs/chromosomes/assembly_1.00/chromosome_defs_v1.00_sorted.txt',
 283 #           abstract=>'test abstract',
 284 #           temp_dir=>$temp_dir,
 285 #           short_name=>'Tomato scaffold map V1.00',
 286 #           long_name=>'Solanum lycopersicum scaffold map V1.00',
 287 #           marker_link => $marker_link,
 288 #                                                   } );
 289 #     }
 290     elsif ($id =~ /^u\d+$/i) {
 291         return CXGN::Cview::Map::SGN::User->new($self->get_dbh(), $id);
 292     }
 293     elsif ($id =~ /pachy/i) {
 294         my $image_dir = $c->config->{'image_path'};
 295         #print STDERR "**** IMAGE DIR = $image_dir\n";
 296         my $map = CXGN::Cview::Map::SGN::Image->new(
 297             $self->get_dbh(),
 298             $id,
 299             $image_dir.'/maps/tomato_pachytene_images/chr1.png',
 300             $image_dir.'/maps/tomato_pachytene_images/chr2.png',
 301             $image_dir.'/maps/tomato_pachytene_images/chr3.png',
 302             $image_dir.'/maps/tomato_pachytene_images/chr4.png',
 303             $image_dir.'/maps/tomato_pachytene_images/chr5.png',
 304             $image_dir.'/maps/tomato_pachytene_images/chr6.png',
 305             $image_dir.'/maps/tomato_pachytene_images/chr7.png',
 306             $image_dir.'/maps/tomato_pachytene_images/chr8.png',
 307             $image_dir.'/maps/tomato_pachytene_images/chr9.png',
 308             $image_dir.'/maps/tomato_pachytene_images/chr10.png',
 309             $image_dir.'/maps/tomato_pachytene_images/chr11.png',
 310             $image_dir.'/maps/tomato_pachytene_images/chr12.png',
 311             );
 312         $map->set_abstract('This map shows the pachytene chromosomes of tomato. It is only for illustrative purposes and does not contain any markers. <br /><br />Images courtesy of Prof. Stephen Stack, Colorado State University.');
 313         $map->set_short_name('Tomato Pachytene Chromosomes');
 314
 315         return $map;
 316     }
 317
 318     elsif ($id =~ /^c\d+$/i) {
 319         my ($gbrowse_fpc) = map $_->fpc_data_sources, $c->enabled_feature('gbrowse2');
 320         my @dbs;
 321         if ($gbrowse_fpc) {
 322             @dbs = $gbrowse_fpc->databases();
 323             @dbs > 1 and die "I can handle only one db!";
 324         } else {
 325             warn "no GBrowse FPC data sources available, cannot open map $id";
 326             return;
 327         }
 328
 329
 330         my $gbrowse_view_link = $gbrowse_fpc->view_url;
 331         return CXGN::Cview::Map::SGN::Contig->new($self->get_dbh(), $id, {
 332             gbrowse_fpc => $gbrowse_fpc,
 333             short_name => $gbrowse_fpc->description,
 334             long_name => '',
 335             temp_dir => $temp_dir,
 336
 337             #marker_link => $gbrowse_fpc->xrefs(),
 338             abstract => $gbrowse_fpc->extended_description."\n". qq{
 339             <p>This overview shows the counts of contigs along the chromosome. Click on any chromosome to view the individual contigs. More information on each contig can be obtained by by clicking on a specific contig.</p>
 340                 <p>Specific contig IDs, including contigs that are not mapped, can be searched on the <a href="$gbrowse_view_link">FPC viewer page</a>.</p>
 341
 342                                                   },
 343
 344                                                   });
 345     }
 346
 347
 348
 349
 350     return;
 351
 352 }
 353
 354 =head2 function get_all_maps()
 355
 356   Synopsis:
 357   Arguments:    none
 358   Returns:      a list of all maps currently defined, as
 359                 CXGN::Cview::Map objects (and subclasses)
 360   Side effects: Queries the database for certain maps
 361   Description:
 362
 363 =cut
 364
 365 sub get_all_maps {
 366     my $self = shift;
 367
 368     my @system_maps = $self->get_system_maps();
 369     my @user_maps = $self->get_user_maps();
 370     my @maps = (@system_maps, @user_maps);
 371     return @maps;
 372
 373 }
 374
 375
 376 =head2 get_system_maps
 377
 378   Usage:        my @system_maps = $map_factory->get_system_maps();
 379   Desc:         retrieves a list of system maps (from the sgn
 380                 database) as a list of CXGN::Cview::Map objects
 381   Ret:
 382   Args:
 383   Side Effects:
 384   Example:
 385
 386 =cut
 387
 388 sub get_system_maps {
 389     my $self = shift;
 390
 391     my @maps = ();
 392
 393     my $query = "SELECT map.map_id FROM sgn.map LEFT JOIN sgn.map_version USING(map_id) LEFT JOIN sgn.accession on(parent_1=accession.accession_id) LEFT JOIN sgn.organism USING(organism_id) LEFT JOIN common_name USING(common_name_id) WHERE current_version='t' ORDER by common_name.common_name";
 394     my $sth = $self->get_dbh()->prepare($query);
 395     $sth->execute();
 396
 397     while (my ($map_id) = $sth->fetchrow_array()) {
 398         my $map = $self->create({ map_id => $map_id });
 399         if ($map) { push @maps, $map; }
 400     }
 401
 402     # push il, physical, contig, and agp map
 403     #
 404     foreach my $id ("il6.5", "il6.9", "p9", "c9", "agp", "pachy") {
 405         my $map = $self->create( {map_id=>$id} );
 406         if ($map) { push @maps, $map; }
 407     }
 408
 409     return @maps;
 410 }
 411
 412
 413
 414 =head2 get_user_maps
 415
 416  Status:       DEPRECATED. Does nothing now, as user maps have been disabled.
 417  Usage:
 418  Desc:         retrieves the current user maps of the logged in user.
 419  Ret:          a list of CXGN::Cview::Map objects
 420  Args:         none
 421  Side Effects: none
 422  Example:
 423
 424 =cut
 425
 426 sub get_user_maps {
 427     my $self = shift;
 428     # push the maps that are specific to that user and not public, if somebody is logged in...
 429     #
 430     my @maps = ();
 431 #     my $login = CXGN::Login->new($self->get_dbh());
 432 #     my $user_id = $login->has_session();
 433 #     if ($user_id) {
 434 #       my $q3 = "SELECT user_map_id FROM sgn_people.user_map WHERE obsolete='f' AND sp_person_id=?";
 435 #       my $h3 = $self->get_dbh()->prepare($q3);
 436 #       $h3->execute($user_id);
 437 #       while (my ($user_map_id) = $h3->fetchrow_array()) {
 438 #           my $map = $self->create( {map_id=>"u".$user_map_id} );
 439
 440 #           if ($map) { push @maps, $map; }
 441 #       }
 442 #     }
 443     return @maps;
 444 }
 445
 446
 447 sub get_db_ids {
 448     my $self = shift;
 449     my $id = shift;
 450
 451     my $population_id = 6;
 452     my $reference_map_id=5;
 453
 454     if ($id=~/il(\d+)\.?(\d*)?/) {
 455         $population_id=$1;
 456         $reference_map_id=$2;
 457     }
 458     if (!$reference_map_id) { $reference_map_id=5; }
 459     if (!$population_id) { $population_id=6; }
 460     #print STDERR "Population ID: $population_id, reference_map_id = $reference_map_id\n";
 461
 462     return ($population_id, $reference_map_id);
 463 }
 464
 465 return 1;