cgi-bin/bulk/download.pl

   1 use CatalystX::GlobalContext qw( $c );
   2 # refactored bulk download script for SGN database
   3 # Alexander Naydich and Matthew Crumb (interns)
   4 # July 3, 2007
   5
   6 ##  based on original bulk download script by
   7 ##  Lukas Mueller
   8 ##  August 12, 2003
   9
  10 ##  Modified August 11, 2005 by
  11 ##  Summer Intern Caroline N. Nyenke
  12
  13 ##  Modified July 7, 2006
  14 ##  Summer Intern Emily Hart
  15
  16 =head1 NAME
  17
  18   /bulk/download.pl
  19
  20 =head1 DESCRIPTION
  21
  22   This perl script is used on the bulk download page. The script collects
  23   identifiers submitted by the user and determines the mode the user is searching
  24   in (clone, unigene, unigene_member, microarray, bac, bac_end, or unigene_convert).
  25   It then uses the appropriate Bulk object to process the input, query the database,
  26   format the results and display them on a separate page. Options of viewing
  27   or downloading in text or fasta format are available.
  28
  29 =cut
  30
  31 use strict;
  32 use warnings;
  33
  34 use CXGN::Page;
  35 use CXGN::DB::Connection;
  36
  37 use CXGN::Bulk::BAC;
  38 use CXGN::Bulk::UnigeneConverter;
  39 use CXGN::Bulk::UnigeneIDUnigene;
  40 use CXGN::Bulk::BACEndRaw;
  41 use CXGN::Bulk::BACEndTrim;
  42 use CXGN::Bulk::CloneEST;
  43 use CXGN::Bulk::CloneUnigene;
  44 use CXGN::Bulk::ArraySpotEST;
  45 use CXGN::Bulk::ArraySpotUnigene;
  46 use CXGN::Bulk::UnigeneMemberInfo;
  47
  48
  49 my $page   = CXGN::Page->new();
  50 my $params = get_parameters($page);
  51
  52 _invalid_params() unless $params->{idType};
  53
  54
  55 $params->{dbc}     = CXGN::DB::Connection->new;
  56 $params->{tempdir} = $c->path_to( $c->tempfiles_subdir('bulk') );
  57
  58 #create correct bulk object
  59 my $bulk;
  60 my $idType = $params->{idType};
  61 if ( $idType eq "bac" ) {
  62     $bulk = CXGN::Bulk::BAC->new($params);
  63 }
  64 elsif ( $idType eq "bac_end" ) {
  65     if ( $params->{bac_seq_type} eq "raw_seq" ) {
  66         $bulk = CXGN::Bulk::BACEndRaw->new($params);
  67     }
  68     elsif ( $params->{bac_seq_type} eq "trim_seq" ) {
  69         $bulk = CXGN::Bulk::BACEndTrim->new($params);
  70     }
  71 }
  72 elsif ( $idType eq "clone" ) {
  73     if ( $params->{seq_type} eq "est_seq" ) {
  74         $bulk = CXGN::Bulk::CloneEST->new($params);
  75     }
  76     elsif ( $params->{seq_type} eq "unigene_seq" ) {
  77         $bulk = CXGN::Bulk::CloneUnigene->new($params);
  78     }
  79 }
  80 elsif ( $idType eq "microarray" ) {
  81     if ( $params->{seq_type} eq "est_seq" ) {
  82         $bulk = CXGN::Bulk::ArraySpotEST->new($params);
  83     }
  84     elsif ( $params->{seq_type} eq "unigene_seq" ) {
  85         $bulk = CXGN::Bulk::ArraySpotUnigene->new($params);
  86     }
  87 }
  88 elsif ( $idType eq "unigene_convert" ) {
  89     $bulk = CXGN::Bulk::UnigeneConverter->new($params);
  90 }
  91 elsif ( $idType eq "unigene" ) {
  92     if ( $params->{unigene_mode} eq "unigene_info" ) {
  93             $bulk = CXGN::Bulk::UnigeneIDUnigene->new($params);
  94     }
  95     elsif ( $params->{unigene_mode} eq "member_info" ) {
  96         $bulk = CXGN::Bulk::UnigeneMemberInfo->new($params);
  97     }
  98 } else {
  99     die "invalid idtype '$idType'";
 100 }
 101
 102 if ( $bulk->process_parameters() ) {
 103     $bulk->process_ids();
 104     #die "process 2 ($bulk)" ;
 105     $bulk->result_summary_page($page);
 106 }
 107 else {
 108     $bulk->error_message();
 109 }
 110 $bulk->clean_up();
 111
 112 exit;
 113
 114 =head2 get_parameters
 115
 116   Desc: sub get_parameters
 117   Args: array; example: get_parameters($ARGV[0]);
 118   Ret : modified parameter array
 119
 120   Retrives parameters from CXGN::Page object (passed from the user).
 121
 122 =cut
 123
 124 sub get_parameters {
 125
 126     my $page = shift;
 127
 128     my $params = {  page => $page };
 129     $params->{idType}               = $page->get_arguments("idType");
 130     $params->{outputType}           = $page->get_arguments("outputType");
 131     $params->{debug}                = $page->get_arguments("debug");
 132     $params->{fasta}                = $page->get_arguments("fasta");
 133     $params->{seq_type}             = $page->get_arguments("seq_type") || '';
 134     $params->{build_id}             = $page->get_arguments("build_id") || '';
 135     $params->{automatic_annotation} = $page->get_arguments("automatic_annotation");
 136     $params->{best_arabidopsis_match} =
 137       $page->get_arguments("best_arabidopsis_match");
 138     $params->{best_genbank_match} = $page->get_arguments("best_genbank_match");
 139     $params->{manual_annotation}  = $page->get_arguments("manual_annotation");
 140
 141     # the file name of the file used in the upload:
 142     $params->{file} = $page->get_arguments("file");
 143
 144     # the name of the dumpfile, if available:
 145     $params->{dumpfile}    = $page->get_arguments("dumpfile");
 146     $params->{page_number} = $page->get_arguments("page_number");
 147
 148     # if this is true, the summary page is displayed again.
 149     $params->{summary} = $page->get_arguments("summary");
 150     ### Build id to deal with:  $params->{build_id}
 151     ### seq_type:$params->{seq_type}
 152     $params->{group_by_unigene} = $page->get_arguments("group_by_unigene");
 153     $params->{unigene_mode}     = $page->get_arguments("unigene_mode");
 154     $params->{associated_loci}  = $page->get_arguments("associated_loci");
 155     $params->{sequence}    = $page->get_arguments("sequence");
 156     $params->{seq_type}    = $page->get_arguments("seq_type");
 157     $params->{unigene_seq} = $page->get_arguments("unigene_seq");
 158     $params->{seq_mode}    = $page->get_arguments("seq_mode");
 159     $params->{est_seq}     = $page->get_arguments("est_seq");
 160     $params->{uni_seq}     = $page->get_arguments("uni_seq");
 161
 162     # bac search parameters
 163     $params->{bac_seq_type}       = $page->get_arguments("bac_seq_type");
 164     $params->{bac_id}             = $page->get_arguments("bac_id");
 165     $params->{bac_end_sequence}   = $page->get_arguments("bac_end_sequence");
 166     $params->{qual_value_seq}     = $page->get_arguments("qual_value_seq");
 167     $params->{arizona_clone_name} = $page->get_arguments("arizona_clone_name");
 168     $params->{chr_clone_name}     = $page->get_arguments("chr_clone_name");
 169     $params->{cornell_clone_name} = $page->get_arguments("cornell_clone_name");
 170     $params->{clone_type}         = $page->get_arguments("clone_type");
 171     $params->{org_name}           = $page->get_arguments("org_name");
 172     $params->{accession_name}     = $page->get_arguments("accession_name");
 173     $params->{library_name}       = $page->get_arguments("library_name");
 174     $params->{estimated_length}   = $page->get_arguments("estimated_length");
 175     $params->{genbank_accession}  = $page->get_arguments("genbank_accession");
 176     $params->{overgo_matches}     = $page->get_arguments("overgo_matches");
 177     $params->{SGN_S}              = $page->get_arguments("SGN_S");
 178     $params->{SGN_C}              = $page->get_arguments("SGN_C");
 179     $params->{SGN_T}              = $page->get_arguments("SGN_T");
 180     $params->{SGN_E}              = $page->get_arguments("SGN_E");
 181     $params->{SGN_U}              = $page->get_arguments("SGN_U");
 182     $params->{chipname}           = $page->get_arguments("chipname");
 183     $params->{TUS}                = $page->get_arguments("TUS");
 184     $params->{clone_name}         = $page->get_arguments("clone_name");
 185     $params->{build_nr}           = $page->get_arguments("build_nr");
 186     $params->{evalue}             = $page->get_arguments("evalue");
 187     $params->{SGN_U_U}            = $page->get_arguments("SGN_U_U");
 188     $params->{SGN_U_M}            = $page->get_arguments("SGN_U_M");
 189     $params->{convert_to_current} = $page->get_arguments("convert_to_current");
 190
 191     # add a new line so that the first id from the file upload is separated
 192     # from the last id here, only if user submits ids in field and as file.
 193
 194     if ( $page->get_arguments("ids") =~ /\S/ ) {
 195         $params->{ids_string} = $page->get_arguments("ids") . "\n";
 196     }
 197
 198     # get an upload object to upload a file
 199     my $upload;
 200     $upload = $page->get_upload();
 201
 202     # if there is a file add it to the ids_string
 203     if ( defined $upload ) {
 204         my $fh = $upload->fh;
 205         if ($fh) {
 206             ### Uploading file...
 207             while (<$fh>) {
 208                 $params->{ids_string} .= $_;
 209             }
 210         }
 211     }
 212
 213     ### size of uploaded ids_string: length $params->{ids_string}
 214
 215     return $params;
 216 }
 217
 218
 219 sub post_only {
 220   my ($page) = @_;
 221   $page->error_page('This page can only accept HTTP POST requests. Please go to <a href="input.pl">Bulk Download</a> to make your selections.');
 222 }
 223
 224 sub _invalid_params {
 225     $c->throw(
 226         title    => "An Error has occured",
 227         message  => "ID Type must be provided",
 228         notify   => 0,
 229         is_error => 0,
 230     );
 231 }
 232
 233 1;
 234
 235 =head1 BUGS
 236
 237   None known.
 238
 239 =head1 AUTHOR
 240
 241   Alexander Naydich and Matthew Crumb (interns)
 242   July 3, 2007
 243
 244   based on original bulk download script by Lukas Mueller
 245   August 12, 2003
 246
 247   Modified August 11, 2005 by
 248   Summer Intern Caroline N. Nyenke
 249
 250   Modified July 7, 2006
 251   Summer Intern Emily Hart
 252
 253 =head1 SEE ALSO
 254
 255   /bulk/display.pl
 256   /bulk/input.pl
 257
 258 =cut
 259