app: allow users to load polyphen/sift, revel and alphamissense scores independently, #TASK-8163

Author: jtarraga

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

@@ -355,6 +355,10 @@ 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_");
     }
@@ -364,7 +368,7 @@ private void loadRevel() throws CellBaseException {
         checkSourceAlreadyLoaded(REVEL_DATA);
 
         HashMap<String, String> collectionMap = new HashMap<>();
-        collectionMap.put(MISSENSE_VARIATION_SCORE_DATA, REVEL_DATA + JSON_GZ_EXTENSION);
+        collectionMap.put(PROTEIN_SUBSTITUTION_PREDICTION_DATA, REVEL_DATA + JSON_GZ_EXTENSION);
 
         Path revelPath = input.resolve(PROTEIN_SUBSTITUTION_PREDICTION_DATA).resolve(REVEL_DATA);
         loadData(revelPath, collectionMap);
@@ -375,7 +379,7 @@ private void loadAlphaMissense() throws CellBaseException {
         checkSourceAlreadyLoaded(ALPHAMISSENSE_DATA);
 
         HashMap<String, String> collectionMap = new HashMap<>();
-        collectionMap.put(MISSENSE_VARIATION_SCORE_DATA, ALPHAMISSENSE_DATA + JSON_GZ_EXTENSION);
+        collectionMap.put(PROTEIN_SUBSTITUTION_PREDICTION_DATA, ALPHAMISSENSE_DATA + JSON_GZ_EXTENSION);
 
         Path alphaMissensePath = input.resolve(PROTEIN_SUBSTITUTION_PREDICTION_DATA).resolve(ALPHAMISSENSE_DATA);
         loadData(alphaMissensePath, collectionMap);

cellbase-lib/src/main/java/org/opencb/cellbase/lib/loader/MongoDBCellBaseLoader.java

@@ -51,7 +51,7 @@
 import java.util.concurrent.BlockingQueue;
 import java.util.stream.Collectors;
 
-import static org.opencb.cellbase.lib.EtlCommons.MISSENSE_VARIATION_SCORE_DATA;
+import static org.opencb.cellbase.lib.EtlCommons.PROTEIN_SUBSTITUTION_PREDICTION_DATA;
 
 /**
  * Created by parce on 18/02/15.
@@ -140,10 +140,10 @@ private String getCollectionName() throws LoaderException {
                     + " are: " + StringUtils.join(releases, ","));
         }
 
-        // Sanity check don't populate collections already populated
-        // Missense variation score data (i.e., revel and alphaMissense) is checked later, since revel and alphamissense are loaded
+        // Sanity check don't populate collections already populated, one exception:
+        // Protein substitution prediction data (i.e., polyphen, sift, revel and alphaMissense) is checked later, since they are loaded
         // in the same collection but independently
-        if (!data.equalsIgnoreCase(MISSENSE_VARIATION_SCORE_DATA)) {
+        if (!data.equalsIgnoreCase(PROTEIN_SUBSTITUTION_PREDICTION_DATA)) {
             for (Release dr : result.getResults()) {
                 if (dr.getRelease() == dataRelease) {
                     if (dr.getCollections().containsKey(data) && dr.getCollections().get(data).equals(collectionName)) {