Tidy up SQL. Add comparison of transcript-probe mappings (array name ignored for now).

misc-scripts/probe_mapping/compare_mappings.pl

@@ -38,22 +38,24 @@ my $t_db = new Bio::EnsEMBL::DBSQL::DBAdaptor('-host'   => $t_host,
 					      '-dbname' => $t_dbname,
 					      '-species'=> $t_dbname);
 
+# TODO fix this when external db is sorted out
+my $restrict_sql = "x.external_db_id > 3000 AND x.external_db_id < 3210";
+
 run();
 
 sub run {
 
   # compare total counts first
-  # TODO fix this when external db is sorted out
-  my $count_sql = "SELECT COUNT(*) FROM xref x, object_xref ox WHERE x.xref_id=ox.xref_id AND x.external_db_id > 3000 AND x.external_db_id < 3210;";
+  my $count_sql = "SELECT COUNT(*) FROM xref x, object_xref ox WHERE x.xref_id=ox.xref_id AND $restrict_sql";
   my $s_sth = $s_db->dbc()->prepare($count_sql);
   $s_sth->execute();
   my $count = ($s_sth->fetchrow_array())[0];
-  print "Total oligo xrefs:\t$s_dbname:\t$count\t";
+  print "Total oligo xrefs in $s_dbname: $count\n";
 
   my $t_sth = $t_db->dbc()->prepare($count_sql);
   $t_sth->execute();
   $count = ($t_sth->fetchrow_array())[0];
-  print "$t_dbname:\t$count\n";
+  print "Total oligo_xrefs in $t_dbname: $count\n";
 
   # cache all mappings from each database
   # key format: transcript_stable_id:array:probe
@@ -75,28 +77,29 @@ sub run {
 
   }
 
+  print "$both mappings in both databases\n";
+  print "$source_only mappings in source only\n";
+
   foreach my $key (keys %target_mappings) {
 
     $target_only++ if (!$source_mappings{$key});
 
   }
 
-  print "$both mappings in both databases\n";
-  print "$source_only mappings in source only\n";
   print "$target_only mappings in target only\n";
 
 }
 
 # ----------------------------------------------------------------------
 
-sub cache_source_mappings {
+sub cache_mappings {
 
   my ($db) = @_;
 
   my %mappings;
 
   # TODO - fix SQL when external_db changes
-  my $sth = $db->dbc()->prepare("SELECT tsi.stable_id, e.db_name, x.dbprimary_acc FROM xref x, external_db e, object_xref ox, transcript_stable_id tsi WHERE x.xref_id=ox.xref_id AND x.external_db_id=e.external_db_id AND e.external_db_id > 3000 AND e.external_db_id < 3210 AND ox.ensembl_object_type='Transcript' AND ox.ensembl_id=tsi.transcript_id");
+  my $sth = $db->dbc()->prepare("SELECT tsi.stable_id, e.db_name, x.dbprimary_acc FROM xref x, external_db e, object_xref ox, transcript_stable_id tsi WHERE x.xref_id=ox.xref_id AND x.external_db_id=e.external_db_id AND ox.ensembl_object_type='Transcript' AND ox.ensembl_id=tsi.transcript_id AND $restrict_sql");
   $sth->execute();
   while (my @row = $sth->fetchrow_array()) {