lib: update loader for Revel data, #TASK-5441, #TASK-5388

opencb · Jan 11, 2024 · 2e764a4 · 2e764a4
1 parent 9229576
commit 2e764a4
Show file tree

Hide file tree

Showing 5 changed files with 13 additions and 11 deletions.
diff --git a/cellbase-app/src/main/java/org/opencb/cellbase/app/cli/admin/AdminCliOptionsParser.java b/cellbase-app/src/main/java/org/opencb/cellbase/app/cli/admin/AdminCliOptionsParser.java
@@ -190,8 +190,8 @@ public class LoadCommandOptions {
         public CommonCommandOptions commonOptions = commonCommandOptions;
 
         @Parameter(names = {"-d", "--data"}, description = "Data model type to be loaded: genome, gene, variation,"
-                + " conservation, regulation, protein, clinical_variants, repeats, regulatory_pfm, splice_score, pubmed, pharmacogenomics."
-                + " 'all' loads everything", required = true, arity = 1)
+                + " conservation, regulation, protein, clinical_variants, repeats, regulatory_pfm, splice_score, pubmed, pharmacogenomics,"
+                + " missense_variation_functional_score, alphamissense; and 'all' loads everything", required = true, arity = 1)
         public String data;
 
         @Parameter(names = {"-i", "--input"}, required = true, arity = 1,

diff --git a/...se-app/src/main/java/org/opencb/cellbase/app/cli/admin/executors/LoadCommandExecutor.java b/...se-app/src/main/java/org/opencb/cellbase/app/cli/admin/executors/LoadCommandExecutor.java
@@ -24,6 +24,7 @@
 import org.opencb.cellbase.core.models.DataRelease;
 import org.opencb.cellbase.core.result.CellBaseDataResult;
 import org.opencb.cellbase.lib.EtlCommons;
+import org.opencb.cellbase.lib.builders.RevelScoreBuilder;
 import org.opencb.cellbase.lib.impl.core.CellBaseDBAdaptor;
 import org.opencb.cellbase.lib.indexer.IndexManager;
 import org.opencb.cellbase.lib.loader.LoadRunner;
@@ -198,16 +199,16 @@ public void execute() throws CellBaseException {
                         }
                         case EtlCommons.MISSENSE_VARIATION_SCORE_DATA: {
                             // Load data
-                            loadIfExists(input.resolve("missense_variation_functional_score.json.gz"),
-                                    "missense_variation_functional_score");
+                            Path path = input.resolve(EtlCommons.PROTEIN_SUBSTITUTION_PREDICTION_DATA);
+                            loadIfExists(path.resolve(EtlCommons.MISSENSE_VARIATION_SCORE_JSON_FILENAME),
+                                    EtlCommons.PROTEIN_SUBSTITUTION_PREDICTION_DATA);
 
                             // Create index
-                            createIndex("missense_variation_functional_score");
+                            createIndex(EtlCommons.PROTEIN_SUBSTITUTION_PREDICTION_DATA);
 
                             // Update release (collection and sources)
-                            List<Path> sources = new ArrayList<>(Collections.singletonList(input.resolve("revelVersion.json")));
-                            dataReleaseManager.update(dataRelease, "missense_variation_functional_score",
-                                    EtlCommons.MISSENSE_VARIATION_SCORE_DATA, sources);
+                            dataReleaseManager.update(dataRelease, EtlCommons.PROTEIN_SUBSTITUTION_PREDICTION_DATA,
+                                    RevelScoreBuilder.SOURCE, Collections.singletonList(path.resolve(EtlCommons.REVEL_VERSION_FILENAME)));
                             break;
                         }
                         case EtlCommons.ALPHAMISSENSE_DATA: {

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/EtlCommons.java
@@ -97,6 +97,8 @@ public class EtlCommons {
     public static final String PROTEIN_FUNCTIONAL_PREDICTION_DATA = "protein_functional_prediction";
     // Revel
     public static final String MISSENSE_VARIATION_SCORE_DATA = "missense_variation_functional_score";
+    public static final String MISSENSE_VARIATION_SCORE_JSON_FILENAME = "missense_variation_functional_score.json.gz";
+    public static final String REVEL_VERSION_FILENAME = "revelVersion.json";
     // AlphaMissense
     public static final String ALPHAMISSENSE_DATA = "alphamissense";
     public static final String ALPHAMISSENSE_RAW_FILENAME = "AlphaMissense_hg38.tsv.gz";

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/RevelScoreBuilder.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/builders/RevelScoreBuilder.java
@@ -34,7 +34,7 @@
 public class RevelScoreBuilder extends CellBaseBuilder {
 
     private Path revelFilePath = null;
-    private static final String SOURCE = "revel";
+    public static final String SOURCE = "revel";
 
     public RevelScoreBuilder(Path revelDirectoryPath, CellBaseSerializer serializer) {
         super(serializer);

diff --git a/cellbase-lib/src/main/java/org/opencb/cellbase/lib/loader/MongoDBCellBaseLoader.java b/cellbase-lib/src/main/java/org/opencb/cellbase/lib/loader/MongoDBCellBaseLoader.java
@@ -151,8 +151,7 @@ private String getCollectionName() throws LoaderException {
                 if (dr.getCollections().containsKey(data)) {
                     String collectionName = CellBaseDBAdaptor.buildCollectionName(data, dataRelease);
                     if (dr.getCollections().get(data).equals(collectionName)) {
-                        throw new LoaderException("Impossible load data " + data + " with release " + dataRelease + " since it"
-                                + " has already been done.");
+                        logger.warn("Loading new data " + data + " with release " + dataRelease + " (already populated previously)");
                     }
                 }
             }