lib/CXGN/Bulk/BACEndTrim.pm

   1 # Bulk BAC End Trim download script for SGN database
   2 # Lukas Mueller, August 12, 2003
   3
   4 # This bulk download option handles the query
   5 # Of BAC Ends of type Trimed.
   6 # Many of its methods are in the Bulk object.
   7
   8 # Modified July 15, 2005
   9 # Modified more August 11, 2005
  10 # Summer Intern Caroline N. Nyenke
  11
  12 # Modified July 7, 2006
  13 # Summer Intern Emily Hart
  14
  15 # Modified July 3rd, 2007
  16 # Alexander Naydich and Matthew Crumb
  17
  18 =head1 NAME
  19
  20   /CXGN/Bulk/BACEndTrim.pm
  21   (A subclass of Bulk)
  22
  23 =head1 DESCRIPTION
  24
  25   This perl script is used on the bulk download page. The script collects
  26   identifiers submitted by the user and returns information based on the
  27   BAC End Trimed Ids entered. It then determines the information the user is
  28   searching for (Bac IDs, Clone Type, Orgonism Name, Accession Name,
  29   Library Name, Estimated Length, Genbank Accessionm, Bac End Sequence,
  30   and Quakl Value Sequence) and preforms the appropriate querying of the
  31   database. The results of the database query are formated and presented
  32   to the user on a separate page. Options of viewing or downloading
  33   in text or fasta are available.
  34
  35 =cut
  36
  37
  38 package CXGN::Bulk::BACEndTrim;
  39 use strict;
  40 use warnings;
  41 use CXGN::Bulk;
  42 use CXGN::Genomic::CloneNameParser;
  43 use CXGN::Genomic::Chromat;
  44 use CXGN::Genomic::GSS;
  45
  46 use CXGN::DB::DBICFactory;
  47
  48 use base "CXGN::Bulk";
  49
  50 sub new
  51 {
  52     my $class = shift;
  53     my $self = $class->SUPER::new(@_);
  54     return $self;
  55 }
  56
  57 =head2 process_parameters
  58
  59   Desc:
  60   Args: none
  61   Ret : 1 if the parameters were OK, 0 if not
  62
  63   Modifies some of the parameters received set in get_parameters. Preparing
  64   data for the database query.
  65
  66 =cut
  67
  68 sub process_parameters
  69 {
  70     my $self = shift;
  71
  72     # @output_list defines the identity on order of all fields that can be output
  73
  74     my @output_list = ('bac_id', 'clone_type', 'org_name',
  75                                 'accession_name', 'library_name', 'estimated_length',
  76                                 'genbank_accession', 'overgo_matches',
  77                                 'bac_end_sequence', 'qual_value_seq');
  78
  79     my @output_fields = ();
  80
  81     $self->debug("Type of identifier: ".($self->{idType})."");
  82
  83     # @output_fields is the sub-set of fields that will actually be output.
  84     foreach my $o (@output_list)
  85     {
  86         if (my $value = $self->{$o})
  87         {
  88             if ($value eq "on")
  89             {
  90             push @output_fields, $o;
  91             }
  92         }
  93     }
  94
  95     $self->{output_list} = \@output_list;
  96     $self->{output_fields} = \@output_fields;
  97
  98     my @ids = $self->check_ids();
  99     if (@ids == ()) {return 0;}
 100     $self->debug("IDs to be processed:");
 101     foreach my $i (@ids)
 102     {
 103         $self->debug($i);
 104     }
 105     my $has_valid_id = 0;
 106     foreach my $i(@ids)
 107     {
 108         if ($i ne "")
 109         {
 110             $has_valid_id = 1;
 111         }
 112     }
 113     if(!$has_valid_id)
 114     {
 115         return 0;
 116     }
 117     $self->{ids} = \@ids;
 118
 119     return 1; #params were OK if we got here
 120 }
 121
 122 =head2 proces_sids
 123
 124   Desc: sub process_[idType]_ids
 125   Args: default;
 126   Ret : data from database printed to a file;
 127
 128   Queries database using Persistent (see perldoc Persistent) and
 129   object oriented perl to obtain data on Bulk Objects using formatted
 130   IDs.
 131
 132 =cut
 133
 134 sub process_ids
 135 {
 136     my $self = shift;
 137     $self->{query_start_time} = time();
 138     my $dbh = $self->{db};
 139     my $chado = CXGN::DB::DBICFactory->open_schema('Bio::Chado::Schema');
 140     my @output_fields = @{$self->{output_fields}};
 141     my @notfound = ();
 142     my @return_data = ();
 143     my ($dump_fh, $notfound_fh) = $self->create_dumpfile();
 144     my @bac_output;
 145     # time counting
 146     my $current_time= time() - $self->{query_start_time};
 147     my $foundcount=0;
 148     my $notfoundcount=0;
 149     my $count=0;
 150
 151     # iterate through identifiers
 152     foreach my $id (@{$self->{ids}}) {
 153       $count++;
 154       my $bac_end_parser = CXGN::Genomic::CloneNameParser->new(); # parse name
 155       my $parsed_bac_end = $bac_end_parser->BAC_end_external_id ($id);
 156
 157       # parsed clone returns undef if parsing did not succeed
 158       unless ($parsed_bac_end) {
 159         print $notfound_fh (">$id\n");
 160         next;
 161       }
 162
 163       #look up the chromat
 164       my $chromat = CXGN::Genomic::Chromat->retrieve($parsed_bac_end->{chromat_id});
 165       unless ($chromat) {
 166         print $notfound_fh (">$id\n");
 167         next;
 168       }
 169
 170       my $clone = $chromat->clone_object;
 171       my $lib = $clone->library_object;
 172       my ($gss) = CXGN::Genomic::GSS->search(chromat_id => $chromat->chromat_id,
 173                                              version    => $parsed_bac_end->{version},
 174                                             );
 175       unless($gss) {
 176         print $notfound_fh ">$id\n";
 177         next;
 178       }
 179
 180       # get organism name and accession
 181       my (undef, $oname, $cname) = $lib->accession_name();
 182
 183
 184       # trim seq and qual value
 185       my $qualvalue = $gss->trimmed_qual;
 186       my $bacseq = $gss->trimmed_seq;
 187
 188 #       # check which parameters were selected
 189 #       my @use_flags = @{$self}{qw/ bac_id
 190 #                                  clone_type
 191 #                                  org_name
 192 #                                  accession_name
 193 #                                  library_name
 194 #                                  estimated_length
 195 #                                  genbank_accession
 196 #                                  overgo_matches
 197 #                                  bac_end_sequence
 198 #                                  qual_value_seq
 199 #                                  /};
 200
 201       # will be added soon
 202       my $bac_id = $chromat->clone_read_external_identifier();
 203       my $clone_type = $parsed_bac_end->{clonetype};
 204       my $library_name = $lib->name();
 205       my $estimated_length = $clone->estimated_length();
 206       my $genbank_accession = $clone->genbank_accession($chado);
 207       my $overgo = "overgo";
 208
 209       my %field_vals = ( "bac_id" => $bac_id,
 210                          "clone_type" => $clone_type,
 211                          "org_name" => $oname,
 212                          "accession_name" => $cname,
 213                          "library_name" => $library_name,
 214                          "estimated_length" => $estimated_length,
 215                          "genbank_accession" =>  $genbank_accession ,
 216                          "overgo_matches" => $overgo,
 217                          "bac_end_sequence" =>  $bacseq,
 218                          "qual_value_seq" => $qualvalue,
 219                                         );
 220
 221       #warn 'made field vals ',join(', ',@field_vals);
 222
 223       my @data_array = ();
 224       foreach my $selected_field (@output_fields) {
 225           push @data_array, $field_vals{$selected_field};
 226       }
 227
 228 #       my @data_array = map { my $val = shift @field_vals;
 229 #                            $_ ? ($val) : ()
 230 #                          } @output_fields;
 231       # warn "information from query: $oname, $cname,\n";
 232
 233       # print query results to dumpfile
 234       print $dump_fh join("\t", @data_array)."\n";
 235
 236
 237     }
 238     $current_time = time() - $self->{query_start_time};
 239     close($dump_fh);
 240     close($notfound_fh);
 241     $self->{foundcount}= $foundcount;
 242     $self->{notfoundcount}= $notfoundcount;
 243     $current_time = time() - $self->{query_start_time};
 244     $self->{query_time} = time() - $self->{query_start_time};
 245 }
 246
 247 1;