lib/SGN/Controller/solGS/Dataset.pm

   1 package SGN::Controller::solGS::Dataset;
   2
   3 use Moose;
   4 use namespace::autoclean;
   5
   6 use Carp qw/ carp confess croak /;
   7 use File::Slurp qw /write_file read_file :edit prepend_file/;
   8 use JSON;
   9 use POSIX qw(strftime);
  10 use Scalar::Util qw /weaken reftype/;
  11 use Storable qw/ nstore retrieve /;
  12 #BEGIN { extends 'Catalyst::Controller' }
  13 use Data::Dumper;
  14
  15 BEGIN { extends 'Catalyst::Controller::REST' }
  16
  17
  18
  19 __PACKAGE__->config(
  20     default   => 'application/json',
  21     stash_key => 'rest',
  22     map       => { 'application/json' => 'JSON',
  23                    'text/html' => 'JSON' },
  24     );
  25
  26
  27
  28 sub get_dataset_trials :Path('/solgs/get/dataset/trials') Args(0) {
  29     my ($self, $c)  = @_;
  30
  31     my $dataset_id = $c->req->param('dataset_id');
  32
  33     croak "Dataset id missing." if !$dataset_id;
  34
  35     $c->stash->{dataset_id} = $dataset_id;
  36     $self->get_dataset_trials_details($c);
  37
  38     $c->stash->{rest}{'trials_ids'} = $c->stash->{trials_ids};
  39     $c->stash->{rest}{'combo_pops_id'} = $c->stash->{combo_pops_id};
  40     $c->stash->{rest}{'trials_names'} = $c->stash->{trials_names};;
  41     $c->stash->{rest}{'genotyping_protocol_id'} = $c->stash->{genotyping_protocol_id};
  42 }
  43
  44
  45 sub check_predicted_dataset_selection :Path('/solgs/check/predicted/dataset/selection') Args(0) {
  46     my ($self, $c) = @_;
  47
  48     my $args = $c->req->param('arguments');
  49
  50     my $json = JSON->new();
  51     $args = $json->decode($args);
  52
  53     my $training_pop_id  = $args->{training_pop_id};
  54     my $selection_pop_id = $args->{selection_pop_id};
  55     $c->stash->{training_traits_ids} = $args->{training_traits_ids};
  56     $c->stash->{genotyping_protocol_id} = $args->{genotyping_protocol_id};
  57
  58     $c->controller('solGS::Download')->selection_prediction_download_urls($c, $training_pop_id, $selection_pop_id);
  59
  60     my $ret->{output} = $c->stash->{selection_prediction_download};
  61
  62     $ret = to_json($ret);
  63
  64     $c->res->content_type('application/json');
  65     $c->res->body($ret);
  66
  67 }
  68
  69
  70 sub get_dataset_trials_ids {
  71     my ($self, $c) = @_;
  72     my $dataset_id = $c->stash->{dataset_id};
  73
  74     my $data = $self->get_model($c)->get_dataset_data($dataset_id);
  75     my $trials_ids = $data->{categories}->{trials};
  76
  77     $c->controller('solGS::combinedTrials')->catalogue_combined_pops($c, $trials_ids);
  78
  79     my $protocol_id = $self->get_dataset_genotyping_protocol($c);
  80
  81     $c->stash->{genotyping_protocol_id} = $protocol_id;
  82     $c->stash->{dataset_trials_ids} = $trials_ids;
  83     $c->stash->{trials_ids} = $trials_ids;
  84
  85 }
  86
  87
  88 sub get_dataset_trials_details {
  89     my ($self, $c) = @_;
  90
  91     $self->get_dataset_trials_ids($c);
  92     $c->controller('solGS::combinedTrials')->process_trials_list_details($c);
  93
  94 }
  95
  96
  97 sub get_dataset_genotypes_genotype_data {
  98     my ($self, $c) = @_;
  99
 100     $self->get_dataset_genotypes_list($c);
 101     $c->controller('solGS::List')->genotypes_list_genotype_file($c);
 102
 103 }
 104
 105
 106 sub get_dataset_genotypes_list {
 107     my ($self, $c, $dataset_id) = @_;
 108
 109     $dataset_id = $c->stash->{dataset_id} if !$dataset_id;
 110
 111     my $genotypes_ids = $self->get_model($c)->get_genotypes_from_dataset($dataset_id);
 112         my $genotypes  = $c->controller('solGS::List')->transform_uniqueids_genotypes($c, $genotypes_ids);
 113     $c->stash->{genotypes_list} = $genotypes;
 114     $c->stash->{genotypes_ids}  = $genotypes_ids;
 115
 116 }
 117
 118 sub submit_dataset_training_data_query {
 119     my ($self, $c) = @_;
 120
 121     my $dataset_id = $c->stash->{dataset_id};
 122
 123     my $data = $self->get_model($c)->get_dataset_data($dataset_id);
 124
 125     my $geno_protocol = $self->get_dataset_genotyping_protocol($c);
 126
 127     my $query_jobs_file;
 128
 129     if (@{$data->{categories}->{plots}})
 130     {
 131         ###### write dataset training data query job function instead...
 132         $c->stash->{plots_names} = $data->{categories}->{plots};
 133         $self->get_dataset_genotypes_list($c);
 134
 135         $c->controller('solGS::List')->get_list_training_data_query_jobs_file($c);
 136         $query_jobs_file = $c->stash->{list_training_data_query_jobs_file};
 137     }
 138     elsif (@{$data->{categories}->{trials}})
 139     {
 140         my $trials = $data->{categories}->{trials};
 141
 142         $c->controller('solGS::AsyncJob')->get_training_pop_data_query_job_args_file($c, $trials, $geno_protocol);
 143         $query_jobs_file  = $c->stash->{training_pop_data_query_job_args_file};
 144     }
 145
 146     $c->stash->{dependent_jobs} = $query_jobs_file;
 147     $c->controller('solGS::AsyncJob')->run_async($c);
 148 }
 149
 150
 151
 152 sub get_dataset_phenotype_data {
 153     my ($self, $c) = @_;
 154
 155     my $dataset_id = $c->stash->{dataset_id};
 156
 157
 158
 159     my $data = $self->get_model($c)->get_dataset_data($dataset_id);
 160
 161     if ($data->{categories}->{plots}->[0])
 162     {
 163         $c->stash->{plots_ids} = $data->{categories}->{plots};
 164
 165         $c->controller('solGS::List')->plots_list_phenotype_file($c);
 166         $c->stash->{phenotype_file} = $c->stash->{plots_list_phenotype_file};
 167     }
 168     elsif ($data->{categories}->{trials}->[0])
 169     {
 170         my $trials = $data->{categories}->{trials};
 171         $c->stash->{pops_ids_list} = $data->{categories}->{trials};
 172         $c->controller('solGS::List')->get_trials_list_pheno_data($c);
 173     }
 174 }
 175
 176
 177 sub create_dataset_pheno_data_query_jobs {
 178     my ($self, $c) = @_;
 179
 180     my $dataset_id = $c->stash->{dataset_id};
 181     my $data = $self->get_model($c)->get_dataset_data($dataset_id);
 182
 183     if ($data->{categories}->{plots}->[0])
 184     {
 185         $c->stash->{plots_ids} = $data->{categories}->{plots};
 186
 187         $c->controller('solGS::List')->plots_list_phenotype_query_job($c);
 188         $c->stash->{dataset_pheno_data_query_jobs} = $c->stash->{plots_list_phenotype_query_job};
 189     }
 190     elsif ($data->{categories}->{trials}->[0])
 191     {
 192         my $trials_ids = $data->{categories}->{trials};
 193
 194         $c->controller('solGS::combinedTrials')->multi_pops_pheno_files($c, $trials_ids);
 195         $c->stash->{phenotype_files_list} = $c->stash->{multi_pops_pheno_files};
 196
 197         $c->controller('solGS::AsyncJob')->get_trials_phenotype_query_jobs_args($c, $trials_ids);
 198         $c->stash->{dataset_pheno_data_query_jobs} = $c->stash->{trials_phenotype_query_jobs_args};
 199     }
 200 }
 201
 202
 203 sub create_dataset_geno_data_query_jobs {
 204     my ($self, $c) = @_;
 205
 206     my $dataset_id = $c->stash->{dataset_id};
 207
 208     my $data = $self->get_model($c)->get_dataset_data($dataset_id);
 209
 210     my $geno_protocol = $self->get_dataset_genotyping_protocol($c);
 211
 212     if ($data->{categories}->{accessions}->[0])
 213     {
 214                $self->dataset_genotype_query_jobs($c);
 215     }
 216     elsif ($data->{categories}->{trials}->[0])
 217     {
 218         my $trials_ids = $data->{categories}->{trials};
 219         $c->controller('solGS::combinedTrials')->multi_pops_geno_files($c, $trials_ids);
 220         $c->stash->{genotype_files_list} = $c->stash->{multi_pops_geno_files};
 221
 222         $c->controller('solGS::AsyncJob')->get_trials_genotype_query_jobs_args($c, $trials_ids, $geno_protocol);
 223         $c->stash->{dataset_geno_data_query_jobs} = $c->stash->{trials_genotype_query_jobs_args};
 224     }
 225 }
 226
 227
 228 sub dataset_genotype_query_jobs {
 229     my ($self, $c) = @_;
 230
 231     my $dataset_id = $c->stash->{dataset_id};
 232     my $protocol_id = $c->stash->{genotyping_protocol_id};
 233
 234         my $pop_id = 'dataset_' . $dataset_id;
 235         my $data_dir =  $c->stash->{solgs_datasets_dir};
 236         my $pop_type = 'dataset';
 237
 238     $c->controller('solGS::Files')->genotype_file_name($c, $pop_id);
 239     my $geno_file = $c->stash->{genotype_file_name};
 240
 241     my $args = {
 242     'dataset_id'=>$dataset_id,
 243         'data_dir'  => $data_dir,
 244         'genotype_file'  => $geno_file,
 245         'genotyping_protocol_id'=> $protocol_id,
 246         'r_temp_file'    => "genotypes-list-genotype-data-query-${pop_id}",
 247     };
 248
 249     $c->stash->{r_temp_file} = $args->{r_temp_file};
 250     $c->controller('solGS::AsyncJob')->create_cluster_accessible_tmp_files($c);
 251     my $out_temp_file = $c->stash->{out_file_temp};
 252     my $err_temp_file = $c->stash->{err_file_temp};
 253
 254     my $temp_dir = $c->stash->{solgs_tempfiles_dir};
 255     my $background_job = $c->stash->{background_job};
 256
 257     my $report_file = $c->controller('solGS::Files')->create_tempfile($temp_dir, "geno-data-query-report-args-${pop_id}");
 258     $c->stash->{report_file} = $report_file;
 259
 260      my $config_args = {
 261         'temp_dir' => $temp_dir,
 262         'out_file' => $out_temp_file,
 263         'err_file' => $err_temp_file,
 264         'cluster_host' => 'localhost'
 265      };
 266
 267     my $config = $c->controller('solGS::AsyncJob')->create_cluster_config($c, $config_args);
 268
 269     my $args_file = $c->controller('solGS::Files')->create_tempfile($temp_dir, "geno-data-query-job-args-file-${pop_id}");
 270
 271     nstore $args, $args_file
 272                 or croak "data query script: $! serializing genotype lists genotype query details to $args_file ";
 273
 274         my $dbhost = $c->config->{dbhost};
 275         my $dbname = $c->config->{dbname};
 276         my $dbpass = $c->config->{dbpass};
 277         my $dbuser = $c->config->{dbuser};
 278
 279     my $cmd = 'mx-run solGS::queryJobs '
 280         . ' --dbhost ' . $dbhost
 281         .' --dbname ' . $dbname
 282         .' --dbuser ' . $dbuser
 283         .' --dbpass ' . $dbpass
 284         . ' --data_type genotype '
 285         . ' --population_type ' . $pop_type
 286         . ' --args_file ' . $args_file;
 287
 288     my $job_args = {
 289         'cmd' => $cmd,
 290         'config' => $config,
 291         'background_job'=> $background_job,
 292         'temp_dir' => $temp_dir,
 293     };
 294
 295     $c->stash->{dataset_geno_data_query_jobs} = $job_args;
 296
 297 }
 298
 299
 300 sub get_dataset_genotyping_protocol {
 301     my ($self, $c, $dataset_id) = @_;
 302
 303     $dataset_id = $c->stash->{dataset_id} if !$dataset_id;
 304
 305     my $data = $self->get_model($c)->get_dataset_data($dataset_id);
 306
 307     my $protocol_id = $data->{categories}->{genotyping_protocols};
 308
 309     if (reftype($protocol_id) eq 'ARRAY')
 310     {
 311         $protocol_id = $protocol_id->[0];
 312     }
 313
 314     $c->controller('solGS::genotypingProtocol')->stash_protocol_id($c, $protocol_id);
 315     $protocol_id = $c->stash->{genotyping_protocol_id};
 316
 317     return $protocol_id;
 318 }
 319
 320
 321 sub get_dataset_plots_list {
 322     my ($self, $c) = @_;
 323
 324     my $dataset_id = $c->stash->{dataset_id};
 325     my $plots = $self->get_model($c)->get_dataset_plots_list($dataset_id);
 326
 327     $c->stash->{plots_names} = $plots;
 328     $c->controller('solGS::List')->get_plots_list_elements_ids($c);
 329
 330 }
 331
 332
 333 sub get_model {
 334     my $self = shift;
 335     my $c = shift;
 336
 337     return $c->controller('solGS::Search')->model($c);
 338
 339 }
 340
 341
 342 sub dataset_population_summary {
 343     my ($self, $c) = @_;
 344
 345     my $dataset_id = $c->stash->{dataset_id};
 346
 347     my $file_id = $self->dataset_file_id($c);
 348     my $tmp_dir = $c->stash->{solgs_datasets_dir};
 349
 350     if (!$c->user)
 351     {
 352         my $page = "/" . $c->req->path;
 353         $c->res->redirect("/solgs/login/message?page=$page");
 354         $c->detach;
 355     }
 356     else
 357     {
 358         my $user_name = $c->user->id;
 359         my $protocol  = $c->controller('solGS::genotypingProtocol')->create_protocol_url($c);
 360
 361         if ($dataset_id)
 362         {
 363             $c->controller('solGS::Files')->population_metadata_file($c, $tmp_dir, $file_id);
 364             my $metadata_file = $c->stash->{population_metadata_file};
 365
 366             my @metadata = read_file($metadata_file, {binmode => ':utf8'});
 367
 368             my ($key, $dataset_name, $desc);
 369
 370             ($desc)        = grep {/description/} @metadata;
 371             ($key, $desc)  = split(/\t/, $desc);
 372
 373             ($dataset_name)       = grep {/dataset_name/} @metadata;
 374             ($key, $dataset_name) = split(/\t/, $dataset_name);
 375
 376             $c->stash(project_id          => $file_id,
 377                       project_name        => $dataset_name,
 378                       selection_pop_name => $dataset_name,
 379                       project_desc        => $desc,
 380                       owner               => $user_name,
 381                       protocol            => $protocol,
 382                 );
 383         }
 384     }
 385 }
 386
 387
 388 sub create_dataset_population_metadata {
 389     my ($self, $c) = @_;
 390
 391     my $dataset_name = $self->get_dataset_name($c);
 392
 393     my $metadata = 'key' . "\t" . 'value';
 394     $metadata .= "\n" . 'user_id' . "\t" . $c->user->id;
 395     $metadata .= "\n" . 'dataset_name' . "\t" . $dataset_name;
 396     $metadata .= "\n" . 'description' . "\t" . 'Uploaded on: ' . strftime "%a %b %e %H:%M %Y", localtime;
 397
 398     $c->stash->{dataset_metadata} = $metadata;
 399
 400 }
 401
 402
 403 sub get_dataset_name {
 404     my ($self, $c, $dataset_id) = @_;
 405
 406     $dataset_id = $c->stash->{dataset_id} if !$dataset_id;
 407     $dataset_id =~ s/\w+_//g;
 408
 409     my $dataset_name = $c->controller('solGS::Search')->model($c)->get_dataset_name($dataset_id);
 410     return $dataset_name;
 411
 412
 413 }
 414
 415
 416 sub create_dataset_population_metadata_file {
 417     my ($self, $c) = @_;
 418
 419     my $file_id = $self->dataset_file_id($c);
 420
 421     my $tmp_dir = $c->stash->{solgs_datasets_dir};
 422
 423     $c->controller('solGS::Files')->population_metadata_file($c,  $tmp_dir, $file_id,);
 424     my $file = $c->stash->{population_metadata_file};
 425
 426     $self->create_dataset_population_metadata($c);
 427     my $metadata = $c->stash->{dataset_metadata};
 428
 429     write_file($file, {binmode => ':utf8'}, $metadata);
 430
 431     $c->stash->{dataset_metadata_file} = $file;
 432
 433 }
 434
 435
 436 sub create_dataset_pop_data_files {
 437     my ($self, $c) = @_;
 438
 439     my $file_id = $self->dataset_file_id($c);
 440     #my $dataset_id = $c->stash->{dataset_id}
 441     $c->controller('solGS::Files')->phenotype_file_name($c, $file_id);
 442     my $pheno_file = $c->stash->{phenotype_file_name};
 443     my $protocol_id = $self->get_dataset_genotyping_protocol($c);
 444     $c->controller('solGS::Files')->genotype_file_name($c, $file_id, $protocol_id);
 445     my $geno_file = $c->stash->{genotype_file_name};
 446
 447     my $files = { pheno_file => $pheno_file, geno_file => $geno_file};
 448
 449     return $files;
 450
 451 }
 452
 453
 454 sub dataset_plots_list_phenotype_file {
 455     my ($self, $c) = @_;
 456
 457     my $dataset_id  = $c->stash->{dataset_id};
 458     my $plots_ids = $c->controller('solGS::Search')->model($c)->get_dataset_plots_list($dataset_id);
 459     my $file_id = $self->dataset_file_id($c);
 460
 461     $c->stash->{pop_id} = $file_id;
 462     $c->controller('solGS::Files')->traits_list_file($c);
 463     my $traits_file =  $c->stash->{traits_list_file};
 464
 465     my $data_dir = $c->stash->{solgs_datasets_dir};
 466
 467     $c->controller('solGS::Files')->phenotype_file_name($c, $file_id);
 468     my $pheno_file = $c->stash->{phenotype_file_name};
 469     #$c->stash->{dataset_plots_list_phenotype_file} = $pheno_file;
 470
 471     $c->controller('solGS::Files')->phenotype_metadata_file($c);
 472     my $metadata_file = $c->stash->{phenotype_metadata_file};
 473
 474     my $args = {
 475         'dataset_id'     => $dataset_id,
 476         'plots_ids'      => $plots_ids,
 477         'traits_file'    => $traits_file,
 478         #'data_dir'       => $data_dir,
 479         'phenotype_file' => $pheno_file,
 480         'metadata_file'  => $metadata_file,
 481         'r_temp_file'    => 'dataset-phenotype-data-query',
 482         'population_type' => 'plots_list'
 483     };
 484
 485     $c->controller('solGS::List')->submit_list_phenotype_data_query($c, $args);
 486     $c->stash->{phenotype_file} = $c->stash->{dataset_plots_list_phenotype_file};
 487
 488 }
 489
 490
 491 sub dataset_file_id {
 492     my ($self, $c) = @_;
 493
 494     my $dataset_id = $c->stash->{dataset_id};
 495     if ( $dataset_id =~ /dataset/) {
 496         return $dataset_id;
 497     }  else {
 498         return 'dataset_' . $dataset_id;
 499     }
 500
 501 }
 502
 503
 504 sub begin : Private {
 505     my ($self, $c) = @_;
 506
 507     $c->controller('solGS::Files')->get_solgs_dirs($c);
 508
 509 }
 510
 511
 512
 513 __PACKAGE__->meta->make_immutable;
 514
 515 ####
 516 1;
 517 ####