lib: update the load command to support PGS data, #TASK-5410, #TASK-5387

opencb · Dec 19, 2023 · 53aa886 · 53aa886
1 parent fd6ac92
commit 53aa886
Show file tree

Hide file tree

Showing 2 changed files with 55 additions and 1 deletion.
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.PolygenicScoreBuilder;
 import org.opencb.cellbase.lib.impl.core.CellBaseDBAdaptor;
 import org.opencb.cellbase.lib.indexer.IndexManager;
 import org.opencb.cellbase.lib.loader.LoadRunner;
@@ -81,7 +82,7 @@ public LoadCommandExecutor(AdminCliOptionsParser.LoadCommandOptions loadCommandO
                     EtlCommons.PROTEIN_FUNCTIONAL_PREDICTION_DATA, EtlCommons.VARIATION_DATA,
                     EtlCommons.VARIATION_FUNCTIONAL_SCORE_DATA, EtlCommons.CLINICAL_VARIANTS_DATA, EtlCommons.REPEATS_DATA,
                     EtlCommons.OBO_DATA, EtlCommons.MISSENSE_VARIATION_SCORE_DATA, EtlCommons.SPLICE_SCORE_DATA, EtlCommons.PUBMED_DATA,
-                    EtlCommons.PHARMACOGENOMICS_DATA};
+                    EtlCommons.PHARMACOGENOMICS_DATA, EtlCommons.PGS_DATA};
         } else {
             loadOptions = loadCommandOptions.data.split(",");
         }
@@ -299,6 +300,11 @@ public void execute() throws CellBaseException {
                             loadPharmacogenomica();
                             break;
                         }
+                        case EtlCommons.PGS_DATA: {
+                            // Load data, create index and update release
+                            loadPolygenicScores();
+                            break;
+                        }
                         default:
                             logger.warn("Not valid 'data'. We should not reach this point");
                             break;
@@ -589,6 +595,49 @@ private void loadPharmacogenomica() throws IOException, CellBaseException {
         dataReleaseManager.update(dataRelease, EtlCommons.PHARMACOGENOMICS_DATA, EtlCommons.PHARMACOGENOMICS_DATA, sources);
     }
 
+    private void loadPolygenicScores() throws NoSuchMethodException, InterruptedException, ExecutionException, InstantiationException,
+            IllegalAccessException, InvocationTargetException, ClassNotFoundException, IOException, CellBaseException, LoaderException {
+        Path pgsPath = input.resolve(EtlCommons.PGS_DATA);
+
+        if (!Files.exists(pgsPath)) {
+            logger.warn("Polygenic scores (PGS) folder {} not found to load", pgsPath);
+            return;
+        }
+
+        // Load common polygenic scores data
+        Path pathToLoad = pgsPath.resolve(PolygenicScoreBuilder.COMMON_POLYGENIC_SCORE_FILENAME);
+        logger.info("Loading file '{}'", pathToLoad.toFile().getName());
+        try {
+            loadRunner.load(pathToLoad, "common_polygenic_score", dataRelease);
+        } catch (ClassNotFoundException | NoSuchMethodException | InstantiationException | InvocationTargetException
+                | IllegalAccessException | ExecutionException | IOException | InterruptedException | CellBaseException
+                | LoaderException e) {
+            logger.error("Error loading file '{}': {}", pathToLoad.toFile().getName(), e.toString());
+        }
+
+        // Load variant polygenic scores data
+        pathToLoad = pgsPath.resolve(PolygenicScoreBuilder.VARIANT_POLYGENIC_SCORE_FILENAME);
+        logger.info("Loading file '{}'", pathToLoad.toFile().getName());
+        try {
+            loadRunner.load(pathToLoad, "variant_polygenic_score", dataRelease);
+        } catch (ClassNotFoundException | NoSuchMethodException | InstantiationException | InvocationTargetException
+                | IllegalAccessException | ExecutionException | IOException | InterruptedException | CellBaseException
+                | LoaderException e) {
+            logger.error("Error loading file '{}': {}", pathToLoad.toFile().getName(), e.toString());
+        }
+
+        // Create index
+        createIndex("variant_polygenic_score");
+        createIndex("common_polygenic_score");
+
+        // Update release (collection and sources)
+        List<Path> sources = new ArrayList<>(Arrays.asList(
+                input.resolve(EtlCommons.PGS_DATA + "/" + EtlCommons.PGS_VERSION_FILENAME)
+        ));
+        dataReleaseManager.update(dataRelease, "variant_polygenic_score", EtlCommons.PGS_DATA, sources);
+        dataReleaseManager.update(dataRelease, "common_polygenic_score", null, null);
+    }
+
     private void createIndex(String collection) {
         if (!createIndexes) {
             return;

diff --git a/cellbase-lib/src/main/resources/mongodb-indexes.json b/cellbase-lib/src/main/resources/mongodb-indexes.json
@@ -145,3 +145,8 @@
 {"collection": "pharmacogenomics", "fields": {"variants.phenotypeType": 1}, "options": {"background": true}}
 {"collection": "pharmacogenomics", "fields": {"variants.confidence": 1}, "options": {"background": true}}
 {"collection": "pharmacogenomics", "fields": {"variants.evidences.pubmed": 1}, "options": {"background": true}}
+
+{"collection": "common_polygenic_score", "fields": {"id": 1}, "options": {"background": true}}
+{"collection": "variant_polygenic_score", "fields": {"_chunkIds": 1}, "options": {"background": true}}
+{"collection": "variant_polygenic_score", "fields": {"chromosome": 1, "position": 1}, "options": {"background": true}}
+{"collection": "variant_polygenic_score", "fields": {"polygenicScores.id": 1}, "options": {"background": true}}