TASK-8163 - Cellbase 6.7: Deleteriousness scores missing

.gitignore

@@ -85,3 +85,4 @@ download_log.json
 
 
 
+.claude

cellbase-app/app/cloud/docker/cellbase-builder/Dockerfile

@@ -1,5 +1,5 @@
-ARG TAG
-FROM opencb/cellbase-base:$TAG
+ARG TAG=latest
+FROM opencb/cellbase-base:${TAG}
 
 LABEL org.label-schema.vendor="OpenCB" \
       org.label-schema.name="cellbase-builder" \
@@ -32,4 +32,4 @@ RUN cd /opt/ensembl && \
 ## Give writting permissions to allow the script ensembl_canonical.pl to create sub-folder for cache purposes
 RUN chmod -R 777 /opt/cellbase/scripts/ensembl-scripts/
 
-ENV PERL5LIB=$PERL5LIB:/opt/ensembl/bioperl-live:/opt/ensembl/ensembl/modules:/opt/ensembl/ensembl-variation/modules:/opt/ensembl/ensembl-funcgen/modules:/opt/ensembl/ensembl-compara/modules:/opt/ensembl/lib/perl/5.18.2:/opt/cellbase/scripts/ensembl-scripts:/opt/ensembl/biomart-perl/lib
+ENV PERL5LIB=/opt/ensembl/bioperl-live:/opt/ensembl/ensembl/modules:/opt/ensembl/ensembl-variation/modules:/opt/ensembl/ensembl-funcgen/modules:/opt/ensembl/ensembl-compara/modules:/opt/ensembl/lib/perl/5.18.2:/opt/cellbase/scripts/ensembl-scripts:/opt/ensembl/biomart-perl/lib

cellbase-app/app/scripts/ensembl-scripts/protein_function_prediction_matrices.pl

@@ -69,15 +69,15 @@
 print "Generating the JSON file for the Sift version.\n";
 $jsonVersion->{"id"} = "sift";
 $jsonVersion->{"name"} = "Sift";
-open(FILE, ">".$outdir."/siftVersion.json") || die "error opening file\n";
+open(FILE, ">".$outdir."/siftVersion.json") || die "error opening file [$outdir/siftVersion.json]: $!\n";
 print FILE to_json($jsonVersion) . "\n";
 close(FILE);
 
 # PolyPhen version file
 print "Generating the JSON file for the PolyPhen version\n";
 $jsonVersion->{"id"} = "polyphen";
 $jsonVersion->{"name"} = "PolyPhen";
-open(FILE, ">".$outdir."/polyphenVersion.json") || die "error opening file\n";
+open(FILE, ">".$outdir."/polyphenVersion.json") || die "error opening file [$outdir/polyphenVersion.json]: $!\n";
 print FILE to_json($jsonVersion) . "\n";
 close(FILE);
 
@@ -158,7 +158,8 @@
 #my @all_chroms = @{$slice_adaptor->fetch_all('chromosome')};
 foreach my $chr(@chromosomes) {
 	my @transcripts = @{$chr->get_all_Transcripts()};
-	open(FILE, ">".$outdir."/prot_func_pred_chr_".$chr->seq_region_name.".json") || die "error opening file\n";
+	my $filename = $outdir."/prot_func_pred_chr_".$chr->seq_region_name.".json";
+	open(FILE, ">".$filename) || die "error opening file [$filename]: $!\n";
 	print @transcripts." transcripts fetched!\n";
 	foreach my $trans(@transcripts) {
 		if($trans->biotype eq 'protein_coding') {

cellbase-app/src/main/java/org/opencb/cellbase/app/cli/admin/executors/LoadCommandExecutor.java

@@ -22,6 +22,7 @@
 import org.opencb.cellbase.app.cli.admin.AdminCliOptionsParser;
 import org.opencb.cellbase.core.config.SpeciesConfiguration;
 import org.opencb.cellbase.core.exception.CellBaseException;
+import org.opencb.cellbase.core.models.DataSource;
 import org.opencb.cellbase.core.models.Release;
 import org.opencb.cellbase.core.result.CellBaseDataResult;
 import org.opencb.cellbase.core.utils.DatabaseNameUtils;
@@ -354,22 +355,34 @@ private void loadConservation() throws IOException, CellBaseException {
     private void loadProteinFunctionalPrediction() throws NoSuchMethodException, InterruptedException, ExecutionException,
             InstantiationException, IllegalAccessException, InvocationTargetException, ClassNotFoundException,
             IOException, CellBaseException, LoaderException {
+        // Check if SIFT/POLYPHEN source has already been loaded
+        checkSourceAlreadyLoaded(SIFT_DATA);
+        checkSourceAlreadyLoaded(POLYPHEN_DATA);
+
         loadData(input.resolve(PROTEIN_SUBSTITUTION_PREDICTION_DATA).resolve(PROTEIN_FUNCTIONAL_PREDICTION_DATA),
                 PROTEIN_SUBSTITUTION_PREDICTION_DATA, "prot_func_pred_");
     }
 
     private void loadRevel() throws CellBaseException {
+        // Check if REVEL source has already been loaded
+        checkSourceAlreadyLoaded(REVEL_DATA);
+
         HashMap<String, String> collectionMap = new HashMap<>();
         collectionMap.put(PROTEIN_SUBSTITUTION_PREDICTION_DATA, REVEL_DATA + JSON_GZ_EXTENSION);
 
-        loadData(input.resolve(PROTEIN_SUBSTITUTION_PREDICTION_DATA).resolve(REVEL_DATA), collectionMap);
+        Path revelPath = input.resolve(PROTEIN_SUBSTITUTION_PREDICTION_DATA).resolve(REVEL_DATA);
+        loadData(revelPath, collectionMap);
     }
 
     private void loadAlphaMissense() throws CellBaseException {
+        // Check if AlphaMissense source has already been loaded
+        checkSourceAlreadyLoaded(ALPHAMISSENSE_DATA);
+
         HashMap<String, String> collectionMap = new HashMap<>();
         collectionMap.put(PROTEIN_SUBSTITUTION_PREDICTION_DATA, ALPHAMISSENSE_DATA + JSON_GZ_EXTENSION);
 
-        loadData(input.resolve(PROTEIN_SUBSTITUTION_PREDICTION_DATA).resolve(ALPHAMISSENSE_DATA), collectionMap);
+        Path alphaMissensePath = input.resolve(PROTEIN_SUBSTITUTION_PREDICTION_DATA).resolve(ALPHAMISSENSE_DATA);
+        loadData(alphaMissensePath, collectionMap);
     }
 
     private void loadClinical() throws FileNotFoundException {
@@ -677,4 +690,16 @@ private Release getDataReleaseForLoading(DataReleaseManager dataReleaseManager)
         }
         return lastDataRelease;
     }
+
+    private void checkSourceAlreadyLoaded(String sourceId) throws CellBaseException {
+        Release release = getDataReleaseForLoading(dataReleaseManager);
+        if (release.getSources() != null) {
+            for (DataSource source : release.getSources()) {
+                if (sourceId.equalsIgnoreCase(source.getId())) {
+                    throw new CellBaseException("Loading data '" + sourceId + "' with release " + dataRelease
+                            + " failed: source '" + sourceId + "' already loaded previously");
+                }
+            }
+        }
+    }
 }