Change regulatory_search_region to have an analysis_id rather than a type.

misc-scripts/regulatory_regions/RegulatoryFeatureParser/BaseParser.pm

@@ -35,9 +35,14 @@ sub delete_existing {
   $sth = $db_adaptor->dbc->prepare("DELETE rfeat FROM regulatory_feature rfeat, analysis a WHERE a.analysis_id=rfeat.analysis_id AND LOWER(a.analysis_id)=?");
   $sth->execute($t);
 
-  # delete search regions of this type
-  $sth = $db_adaptor->dbc->prepare("DELETE FROM regulatory_search_region WHERE type=?");
-  $sth->execute($t);
+  # Delete search regions; they have a different analysis_id
+  my $sr_type = $type . "_search";
+  die "Can't find analysis for $sr_type " unless validate_type($db_adaptor, $sr_type);
+  my $anal_sth = $db_adaptor->dbc->prepare("SELECT analysis_id FROM analysis WHERE LOWER(logic_name)=?");
+  $anal_sth->execute($sr_type);
+  my $anal = ($anal_sth->fetchrow_array())[0];
+  $sth = $db_adaptor->dbc->prepare("DELETE FROM regulatory_search_region WHERE analysis_id=?");
+  $sth->execute($anal);
 
 }
 
@@ -127,7 +132,7 @@ sub upload_features_and_factors {
   my $factor_sth = $dbc->prepare("INSERT INTO regulatory_factor (regulatory_factor_id, name, type) VALUES(?,?,?)");
   my $feature_object_sth = $dbc->prepare("INSERT INTO regulatory_feature_object (regulatory_feature_id, ensembl_object_type, ensembl_object_id, influence, evidence) VALUES(?,?,?,?,?)");
 
-  my $sr_sth = $dbc->prepare("INSERT INTO regulatory_search_region (name, seq_region_id, seq_region_start, seq_region_end, seq_region_strand, ensembl_object_type, ensembl_object_id, type ) VALUES(?,?,?,?,?,?,?,?)");
+  my $sr_sth = $dbc->prepare("INSERT INTO regulatory_search_region (name, seq_region_id, seq_region_start, seq_region_end, seq_region_strand, ensembl_object_type, ensembl_object_id, analysis_id ) VALUES(?,?,?,?,?,?,?,?)");
 
   print "Uploading " . scalar(@{$objects->{FEATURES}}) . " features ...\n";
 
@@ -171,7 +176,7 @@ sub upload_features_and_factors {
 		     $search_region->{STRAND},
 		     $search_region->{ENSEMBL_OBJECT_TYPE},
 		     $search_region->{ENSEMBL_OBJECT_ID},
-		     $search_region->{TYPE});
+		     $search_region->{ANALYSIS_ID});
 
 
   }

misc-scripts/regulatory_regions/RegulatoryFeatureParser/cisred.pm

@@ -103,16 +103,19 @@ sub parse {
   close(GROUP_SIZES);
 
   # ----------------------------------------
-  # Analysis
+  # Analysis - need one for each type of feature
+  my %analysis;
 
-  my $analysis = $analysis_adaptor->fetch_by_logic_name("cisRed");
+  foreach my $anal ("cisRed", "cisred_search") {  # TODO - add other types as necessary
 
-  if (!$analysis) {
-    print STDERR "Can't get analysis for cisRed, skipping\n";
-    next;
-  }
+    my $analysis_obj = $analysis_adaptor->fetch_by_logic_name($anal);
+
+    die "Can't get analysis for $anal, skipping" if (!$analysis_obj);
 
-  my $analysis_id = $analysis->dbID();
+    $analysis{$anal} = $analysis_obj->dbID();
+    print "Analysis ID for $anal is " . $analysis{$anal} . "\n";
+
+  }
 
   # ----------------------------------------
   # Parse motifs.txt file
@@ -136,7 +139,7 @@ sub parse {
 
     $feature{NAME} = "craHsap" . $motif_id;
     $feature{INFLUENCE} = "unknown"; # TODO - what does cisRed store?
-    $feature{ANALYSIS_ID} = $analysis_id;
+    $feature{ANALYSIS_ID} = $analysis{cisRed};
 
     # ----------------------------------------
     # Factor
@@ -224,7 +227,7 @@ sub parse {
 
   close FILE;
 
- # ----------------------------------------
+  # ----------------------------------------
   # Search regions 
   # read search_regions.txt from same location as $file
   my $search_regions_file = dirname($file) . "/search_regions.txt";
@@ -258,7 +261,7 @@ sub parse {
     $search_region{STRAND} = ($strand =~ /\+/ ? 1 : -1);
     $search_region{ENSEMBL_OBJECT_TYPE} = 'Gene';
     $search_region{ENSEMBL_OBJECT_ID} = $gene_id;
-    $search_region{TYPE} = 'cisred';
+    $search_region{ANALYSIS_ID} = $analysis{cisred_search};
     push @search_regions, \%search_region;
 
   }

sql/table.sql

@@ -1310,7 +1310,7 @@ CREATE TABLE regulatory_search_region (
   seq_region_strand            TINYINT NOT NULL,
   ensembl_object_type          ENUM( 'Transcript', 'Translation', 'Gene') NOT NULL,
   ensembl_object_id            INT,           # FK to gene/transcript/translation
-  type                         ENUM('cisred_search', 'cisred_mask', 'unknown'),
+  analysis_id                  INT NOT NULL,  # FK to analysis
 
   PRIMARY KEY (regulatory_search_region_id),
   KEY rsr_idx (regulatory_search_region_id),