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 CXGN::Cview::Map::SGN::Genetic;
  63 #use CXGN::Cview::Map::SGN::User;
  64 use CXGN::Cview::Map::SGN::Fish;
  65 use CXGN::Cview::Map::SGN::Sequence;
  66 use CXGN::Cview::Map::SGN::IL;
  67 use CXGN::Cview::Map::SGN::Physical;
  68 use CXGN::Cview::Map::SGN::ProjectStats;
  69 use CXGN::Cview::Map::SGN::AGP;
  70 use CXGN::Cview::Map::SGN::ITAG;
  71 use CXGN::Cview::Map::SGN::Contig;
  72 use CXGN::Cview::Map::SGN::Scaffold;
  73
  74 =head2 function new()
  75
  76   Synopsis:     constructor
  77   Arguments:    a database handle
  78   Returns:      a CXGN::Cview::MapFactory::SGN object
  79   Side effects: none
  80   Description:  none
  81
  82 =cut
  83
  84 sub new {
  85     my $class = shift;
  86     my $dbh = shift;
  87     my $context = shift;
  88
  89     if (!$context) {
  90         require SGN::Context;
  91         $context = SGN::Context->new();
  92     }
  93     my $self = $class->SUPER::new($dbh);
  94
  95     $self->{context}=$context;
  96
  97
  98     return $self;
  99 }
 100
 101 =head2 function create()
 102
 103   Description:  creates a map based on the hashref given, which
 104                 should either contain the key map_id or map_version_id
 105                 and an appropriate identifier. The function returns undef
 106                 if a map of the given id cannot be found/created.
 107   Example:
 108
 109 =cut
 110
 111 sub create {
 112     my $self = shift;
 113     my $hashref = shift;
 114     #print STDERR "Hashref = map_id => $hashref->{map_id}, map_version_id => $hashref->{map_version_id}\n";
 115
 116     my $c = $self->{context};
 117     my $temp_dir = $c->path_to( $c->tempfiles_subdir('cview') );
 118
 119     if (!exists($hashref->{map_id}) && !exists($hashref->{map_version_id})) {
 120         die "[CXGN::Cview::MapFactory] Need either a map_id or map_version_id.\n";
 121     }
 122     if ($hashref->{map_id} && $hashref->{map_version_id}) {
 123         die "[CXGN::Cview::MapFactory] Need either a map_id or map_version_id - not both.\n";
 124     }
 125     if ($hashref->{map_id}) {
 126         $hashref->{map_version_id}=CXGN::Cview::Map::Tools::find_current_version($self->get_dbh(), $hashref->{map_id});
 127     }
 128
 129     # now, we only deal with map_versions...
 130     #
 131     my $id = $hashref->{map_version_id};
 132
 133     #print STDERR "MapFactory: dealing with id = $id\n";
 134
 135     # if the map_version_id is purely numeric,
 136     # check if the map is in the maps table and generate the
 137     # appropriate map
 138
 139     if ($id=~/^\d+$/) {
 140         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=?";
 141         my $sth = $self->get_dbh()->prepare($query);
 142         $sth->execute($id);
 143         my ($id, $map_type) = $sth->fetchrow_array();
 144         if ($map_type =~ /genetic/i) {
 145             return CXGN::Cview::Map::SGN::Genetic->new($self->get_dbh(), $id);
 146         }
 147         elsif ($map_type =~ /fish/) {
 148             #print STDERR "Creating a fish map...\n";
 149             return CXGN::Cview::Map::SGN::Fish->new($self->get_dbh(), $id);
 150         }
 151         elsif ($map_type =~ /seq/) {
 152             #print STDERR "Creating a seq map...\n";
 153             return CXGN::Cview::Map::SGN::Sequence->new($self->get_dbh(), $id);
 154         }
 155     }
 156     elsif ($id =~ /^u/i) {
 157         #return CXGN::Cview::Map::SGN::User->new($self->get_dbh(), $id);
 158     }
 159     elsif ($id =~ /^il/i) {
 160         my $abstract =
 161         "The tomato Introgression lines (ILs) are a set of nearly isogenic lines (NILs) developed by Dani Zamir through a succession of backcrosses, where each line carries a single genetically defined chromosome segment from a divergent genome. The ILs, representing whole-genome coverage of S. pennellii in overlapping segments in the genetic background of S. lycopersicum cv. M82, were first phenotyped in 1993, and presently this library consists of 76 genotypes. ";
 162
 163         my ($population_id, $map_id) = $self->get_db_ids($id);
 164         if ($map_id == 9) {
 165             $abstract .= " This IL map is based on markers of the F2-2000 map. ILs have also been mapped <a href=\"map.pl?map_id=il6.5&amp;show_ruler=1&amp;show_offsets=1\" >with the ExPEN1992 map as a reference</a>.";
 166         }
 167         elsif ($map_id ==5) {
 168             $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>. ";
 169         }
 170
 171         my $ref_map = "ExPEN2000";
 172         if ($id==5) { $ref_map = "ExPEN1992";}
 173         my $long_name =qq | <i>Solanum lycopersicum</i> Zamir Introgression Lines (IL) based on $ref_map |;
 174
 175         return CXGN::Cview::Map::SGN::IL->new($self->get_dbh(), $id,
 176                                               { short_name => "Tomato IL map",
 177                                                 long_name => $long_name,
 178                                                 abstract => $abstract
 179                                               });
 180     }
 181     elsif ($id =~ /^\//) {
 182         #return CXGN::Cview::Map::SGN::File->new($dbh, $id);
 183     }
 184     elsif ($id =~ /^p\d+/) {
 185         return CXGN::Cview::Map::SGN::Physical->new($self->get_dbh(), $id);
 186     }
 187     elsif ($id =~ /^o$/i) {
 188
 189         return CXGN::Cview::Map::SGN::ProjectStats->new($self->get_dbh(), {
 190             short_name=>"Tomato Sequencing Progress",
 191             long_name=>"Tomato Sequencing Statistics by Chromosome",
 192             abstract => $self->get_abstract(),
 193
 194         });
 195     }
 196
 197     elsif ($id =~ /^agp$/i) {
 198         return CXGN::Cview::Map::SGN::AGP->new($self->get_dbh(), $id, {
 199                                                short_name => "Tomato AGP map",
 200                                                long_name => "Tomato (Solanum lycopersicum) Accessioned Golden Path map",
 201                                                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>" ,
 202                                                temp_dir => $temp_dir } );
 203     }
 204     elsif ($id =~ /^itag$/i) {
 205
 206         my (@sources) = map $_->data_sources(), $c->enabled_feature('gbrowse2');
 207         my ($gbrowse_itag) = grep $_->description()=~/ITAG_devel.+genomic/i, @sources;
 208         my @dbs;
 209         if ($gbrowse_itag) {
 210             @dbs = $gbrowse_itag->databases();
 211             @dbs > 1 and die "I can handle only one db!";
 212         }
 213
 214         my $gbrowse_view_link = $gbrowse_itag->view_url;
 215
 216         my $marker_link =  sub { my $id = shift; return "$gbrowse_view_link?name=$id"; };
 217         return CXGN::Cview::Map::SGN::ITAG->new($self->get_dbh(), $id, {
 218                                                 short_name => "Tomato ITAG map",
 219                                                 long_name=>"Tomato (Solanum lycopersicum) ITAG map",
 220                                                 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.",
 221                                                 temp_dir => $temp_dir,
 222                                                 marker_link => $marker_link,
 223                                             }
 224             );
 225     }
 226
 227     elsif ($id =~ /scaffold103/) {
 228
 229         return CXGN::Cview::Map::SGN::Scaffold->new($self->get_dbh(), $id, {
 230             file=> '/data/prod/public/tomato_genome/wgs/chromosomes/assembly_1.03/chromosome_defs_v1.03_sorted.txt',
 231             abstract=>'test abstract',
 232             temp_dir=>$temp_dir,
 233             short_name=>'Tomato scaffold map V1.03',
 234             long_name=>'Solanum lycopersicum scaffold map V1.03',
 235             marker_link => sub {},
 236                                                     } );
 237     }
 238
 239     elsif ($id =~ /scaffold100/) {
 240         my (@sources) = map $_->data_sources(), $c->enabled_feature('gbrowse2');
 241         my ($gbrowse) = grep $_->description()=~/ITAG1.+genomic/i, @sources;
 242         if (!$gbrowse) { die "No such map in GBrowse."; }
 243         my @dbs;
 244         if ($gbrowse) {
 245             @dbs = $gbrowse->databases();
 246             @dbs > 1 and die "I can handle only one db!";
 247         }
 248
 249         my $gbrowse_view_link = $gbrowse->view_url;
 250
 251         my $marker_link =  sub { my $id = shift; return "$gbrowse_view_link?name=$id"; };
 252
 253         return CXGN::Cview::Map::SGN::Scaffold->new($self->get_dbh(), $id, {
 254             file=> '/data/prod/public/tomato_genome/wgs/chromosomes/assembly_1.00/chromosome_defs_v1.00_sorted.txt',
 255             abstract=>'test abstract',
 256             temp_dir=>$temp_dir,
 257             short_name=>'Tomato scaffold map V1.00',
 258             long_name=>'Solanum lycopersicum scaffold map V1.00',
 259             marker_link => $marker_link,
 260                                                     } );
 261     }
 262     elsif ($id =~ /^u\d+$/i) {
 263         return CXGN::Cview::Map::SGN::User->new($self->get_dbh(), $id);
 264     }
 265     elsif ($id =~ /^c\d+$/i) {
 266         my ($gbrowse_fpc) = map $_->fpc_data_sources, $c->enabled_feature('gbrowse2');
 267         my @dbs;
 268         if ($gbrowse_fpc) {
 269             @dbs = $gbrowse_fpc->databases();
 270             @dbs > 1 and die "I can handle only one db!";
 271         }
 272
 273         my $gbrowse_view_link = $gbrowse_fpc->view_url;
 274         return CXGN::Cview::Map::SGN::Contig->new($self->get_dbh(), $id, {
 275             gbrowse_fpc => $gbrowse_fpc,
 276             short_name => $gbrowse_fpc->description,
 277             long_name => '',
 278             temp_dir => $temp_dir,
 279             #marker_link => $gbrowse_fpc->xrefs(),
 280             abstract => $gbrowse_fpc->extended_description."\n". qq{
 281             <p>This overview shows the counts of contigs along the chromosome. Click on any chromosome to view the individual contigs. More information on each contig can be obtained by by clicking on a specific contig.</p>
 282                 <p>Specific contig IDs, including contigs that are not mapped, can be searched on the <a href="$gbrowse_view_link">FPC viewer page</a>.</p>
 283
 284                                                   }
 285
 286                                                   });
 287     }
 288
 289
 290
 291
 292     return;
 293
 294 }
 295 =head2 function get_all_maps()
 296
 297   Synopsis:
 298   Arguments:    none
 299   Returns:      a list of all maps currently defined, as
 300                 CXGN::Cview::Map objects (and subclasses)
 301   Side effects: Queries the database for certain maps
 302   Description:
 303
 304 =cut
 305
 306 sub get_all_maps {
 307     my $self = shift;
 308
 309     my @system_maps = $self->get_system_maps();
 310     my @user_maps = $self->get_user_maps();
 311     my @maps = (@system_maps, @user_maps);
 312     return @maps;
 313
 314 }
 315
 316
 317 =head2 get_system_maps
 318
 319   Usage:        my @system_maps = $map_factory->get_system_maps();
 320   Desc:         retrieves a list of system maps (from the sgn
 321                 database) as a list of CXGN::Cview::Map objects
 322   Ret:
 323   Args:
 324   Side Effects:
 325   Example:
 326
 327 =cut
 328
 329 sub get_system_maps {
 330     my $self = shift;
 331
 332     my @maps = ();
 333
 334     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";
 335     my $sth = $self->get_dbh()->prepare($query);
 336     $sth->execute();
 337
 338     while (my ($map_id) = $sth->fetchrow_array()) {
 339         my $map = $self->create({ map_id => $map_id });
 340         if ($map) { push @maps, $map; }
 341     }
 342
 343     # push il, physical, contig, and agp map
 344     #
 345     foreach my $id ("il6.5", "il6.9", "p9", "c9", "agp", "itag") {
 346         my $map = $self->create( {map_id=>$id} );
 347         if ($map) { push @maps, $map; }
 348     }
 349
 350     return @maps;
 351 }
 352
 353
 354
 355 =head2 get_user_maps
 356
 357  Status:       DEPRECATED. Does nothing now, as user maps have been disabled.
 358  Usage:
 359  Desc:         retrieves the current user maps of the logged in user.
 360  Ret:          a list of CXGN::Cview::Map objects
 361  Args:         none
 362  Side Effects: none
 363  Example:
 364
 365 =cut
 366
 367 sub get_user_maps {
 368     my $self = shift;
 369     # push the maps that are specific to that user and not public, if somebody is logged in...
 370     #
 371     my @maps = ();
 372 #     my $login = CXGN::Login->new($self->get_dbh());
 373 #     my $user_id = $login->has_session();
 374 #     if ($user_id) {
 375 #       my $q3 = "SELECT user_map_id FROM sgn_people.user_map WHERE obsolete='f' AND sp_person_id=?";
 376 #       my $h3 = $self->get_dbh()->prepare($q3);
 377 #       $h3->execute($user_id);
 378 #       while (my ($user_map_id) = $h3->fetchrow_array()) {
 379 #           my $map = $self->create( {map_id=>"u".$user_map_id} );
 380
 381 #           if ($map) { push @maps, $map; }
 382 #       }
 383 #     }
 384     return @maps;
 385 }
 386
 387
 388 sub get_db_ids {
 389     my $self = shift;
 390     my $id = shift;
 391
 392     my $population_id = 6;
 393     my $reference_map_id=5;
 394
 395     if ($id=~/il(\d+)\.?(\d*)?/) {
 396         $population_id=$1;
 397         $reference_map_id=$2;
 398     }
 399     if (!$reference_map_id) { $reference_map_id=5; }
 400     if (!$population_id) { $population_id=6; }
 401     #print STDERR "Population ID: $population_id, reference_map_id = $reference_map_id\n";
 402
 403     return ($population_id, $reference_map_id);
 404 }
 405
 406 return 1;