TASK-7142 - OOME running calculate_pedigree_graph

opencga-analysis/src/main/java/org/opencb/opencga/analysis/family/PedigreeGraphInitAnalysis.java

@@ -22,18 +22,18 @@
 import org.opencb.commons.exec.Command;
 import org.opencb.commons.utils.DockerUtils;
 import org.opencb.opencga.analysis.tools.OpenCgaToolScopeStudy;
+import org.opencb.opencga.catalog.db.api.DBIterator;
 import org.opencb.opencga.catalog.exceptions.CatalogException;
+import org.opencb.opencga.catalog.managers.FamilyManager;
 import org.opencb.opencga.catalog.utils.PedigreeGraphUtils;
 import org.opencb.opencga.core.api.ParamConstants;
 import org.opencb.opencga.core.exceptions.ToolException;
 import org.opencb.opencga.core.models.common.Enums;
 import org.opencb.opencga.core.models.family.Family;
-import org.opencb.opencga.core.response.OpenCGAResult;
 import org.opencb.opencga.core.tools.annotations.Tool;
 
 import java.util.Arrays;
 import java.util.List;
-import java.util.stream.Collectors;
 
 @Tool(id = PedigreeGraphInitAnalysis.ID, resource = Enums.Resource.FAMILY)
 public class PedigreeGraphInitAnalysis extends OpenCgaToolScopeStudy {
@@ -68,30 +68,27 @@ protected void run() throws ToolException {
 
         step(getId(), () -> {
             // Get all families from that study
-            QueryOptions queryOptions = new QueryOptions(QueryOptions.INCLUDE, Arrays.asList("id", "members", "pedigreeGraph"));
-            OpenCGAResult<Family> results = catalogManager.getFamilyManager().search(study, new Query(), queryOptions, token);
-
-            // Get families to update by filtering
-            List<Family> familiesToUpdate = results.getResults().stream().filter(family -> PedigreeGraphUtils.hasMinTwoGenerations(family)
-                            && (family.getPedigreeGraph() == null || StringUtils.isEmpty(family.getPedigreeGraph().getBase64())))
-                    .collect(Collectors.toList());
-
-            // Update those families
-            for (Family family: familiesToUpdate) {
-                try {
-                    logger.info("Updating pedigree graph for family '{}'", family.getId());
-                    catalogManager.getFamilyManager().update(study, family.getId(), null,
-                            new QueryOptions(ParamConstants.FAMILY_UPDATE_PEDIGREEE_GRAPH_PARAM, true), token);
-                    String msg = "Updated pedigree graph for family '" + family.getId() + "'";
-                    logger.info(msg);
-                    addInfo(msg);
-                } catch (CatalogException e) {
-                    String msg = "Something wrong happened when updating pedigree graph for family '" + family.getId() + "'. Error: "
-                            + e.getMessage();
-                    logger.info(msg);
-                    addWarning(msg);
+            try (DBIterator<Family> iterator = catalogManager.getFamilyManager().iterator(study, new Query(), FamilyManager.INCLUDE_FAMILY_FOR_PEDIGREE, token)) {
+                while (iterator.hasNext()) {
+                    Family family = iterator.next();
+                    if (PedigreeGraphUtils.hasMinTwoGenerations(family)
+                            && (family.getPedigreeGraph() == null || StringUtils.isEmpty(family.getPedigreeGraph().getBase64()))) {
+                        try {
+                            logger.info("Updating pedigree graph for family '{}'", family.getId());
+                            catalogManager.getFamilyManager().update(study, family.getId(), null,
+                                    new QueryOptions(ParamConstants.FAMILY_UPDATE_PEDIGREEE_GRAPH_PARAM, true), token);
+                            String msg = "Updated pedigree graph for family '" + family.getId() + "'";
+                            logger.info(msg);
+                            addInfo(msg);
+                        } catch (CatalogException e) {
+                            String msg = "Something wrong happened when updating pedigree graph for family '" + family.getId() + "'.";
+                            logger.warn(msg, e);
+                            addWarning(msg + " Error: " + e.getMessage());
+                        }
+                    }
                 }
             }
+
             logger.info("Finished updating pedigree graph for families of the study '{}'", study);
         });
     }

opencga-app/src/main/java/org/opencb/opencga/app/migrations/v2_8_0/catalog/CalculatePedigreeGraphMigration.java

@@ -6,7 +6,9 @@
 import org.opencb.commons.datastore.core.Query;
 import org.opencb.commons.datastore.core.QueryOptions;
 import org.opencb.opencga.analysis.family.PedigreeGraphInitAnalysis;
+import org.opencb.opencga.catalog.db.api.DBIterator;
 import org.opencb.opencga.catalog.exceptions.CatalogException;
+import org.opencb.opencga.catalog.managers.FamilyManager;
 import org.opencb.opencga.catalog.migration.Migration;
 import org.opencb.opencga.catalog.migration.MigrationRun;
 import org.opencb.opencga.catalog.migration.MigrationTool;
@@ -84,16 +86,18 @@ protected void run() throws Exception {
     public List<String> getStudies() throws CatalogException {
         Set<String> studies = new LinkedHashSet<>();
         QueryOptions projectOptions = new QueryOptions(QueryOptions.INCLUDE, Arrays.asList("id", "studies"));
-        QueryOptions familyOptions = new QueryOptions(QueryOptions.INCLUDE, Arrays.asList("id", "members", "pedigreeGraph"));
         for (Project project : catalogManager.getProjectManager().search(new Query(), projectOptions, token).getResults()) {
             if (CollectionUtils.isNotEmpty(project.getStudies())) {
                 for (Study study : project.getStudies()) {
                     String id = study.getFqn();
-                    for (Family family : catalogManager.getFamilyManager().search(id, new Query(), familyOptions, token).getResults()) {
-                        if (PedigreeGraphUtils.hasMinTwoGenerations(family)
-                                && (family.getPedigreeGraph() == null || StringUtils.isEmpty(family.getPedigreeGraph().getBase64()))) {
-                            studies.add(id);
-                            break;
+                    try (DBIterator<Family> iterator = catalogManager.getFamilyManager().iterator(id, new Query(), FamilyManager.INCLUDE_FAMILY_FOR_PEDIGREE, token)) {
+                        while (iterator.hasNext()) {
+                            Family family = iterator.next();
+                            if (PedigreeGraphUtils.hasMinTwoGenerations(family)
+                                    && (family.getPedigreeGraph() == null || StringUtils.isEmpty(family.getPedigreeGraph().getBase64()))) {
+                                studies.add(id);
+                                break;
+                            }
                         }
                     }
                 }

opencga-catalog/src/main/java/org/opencb/opencga/catalog/db/mongodb/FamilyMongoDBAdaptor.java

@@ -38,6 +38,7 @@
 import org.opencb.opencga.catalog.exceptions.CatalogDBException;
 import org.opencb.opencga.catalog.exceptions.CatalogException;
 import org.opencb.opencga.catalog.exceptions.CatalogParameterException;
+import org.opencb.opencga.catalog.managers.FamilyManager;
 import org.opencb.opencga.catalog.managers.IndividualManager;
 import org.opencb.opencga.catalog.utils.Constants;
 import org.opencb.opencga.catalog.utils.ParamUtils;
@@ -543,7 +544,7 @@ private PedigreeGraph computePedigreeGraph(ClientSession clientSession, Family f
         Query query = new Query()
                 .append(QueryParams.UID.key(), family.getUid())
                 .append(QueryParams.STUDY_UID.key(), family.getStudyUid());
-        Family tmpFamily = get(clientSession, query, QueryOptions.empty()).first();
+        Family tmpFamily = get(clientSession, query, FamilyManager.INCLUDE_FAMILY_FOR_PEDIGREE).first();
 
         try {
             return PedigreeGraphUtils.getPedigreeGraph(tmpFamily,

opencga-catalog/src/main/java/org/opencb/opencga/catalog/db/mongodb/IndividualMongoDBAdaptor.java

@@ -1304,7 +1304,8 @@ private MongoDBIterator<Document> getMongoCursor(ClientSession clientSession, Qu
             qOptions = new QueryOptions();
         }
 
-        qOptions = removeInnerProjections(qOptions, QueryParams.SAMPLES.key());
+        qOptions = removeInnerProjections(qOptions, Arrays.asList(QueryParams.SAMPLES.key(), QueryParams.FATHER.key(),
+                QueryParams.MOTHER.key()));
         qOptions = removeAnnotationProjectionOptions(qOptions);
 
         // FIXME we should be able to remove this now safely

opencga-catalog/src/main/java/org/opencb/opencga/catalog/db/mongodb/MongoDBAdaptor.java

@@ -558,42 +558,70 @@ protected QueryOptions extractNestedOptions(QueryOptions options, String key) {
      * @return new QueryOptions after removing the inner projectionKey projections.
      */
     protected QueryOptions removeInnerProjections(QueryOptions options, String projectionKey) {
+        return removeInnerProjections(options, Collections.singletonList(projectionKey));
+    }
+
+    /**
+     * Removes any other entity projections made. This method should be called by any entity containing inner entities:
+     * Family -> Individual; Individual -> Sample; File -> Sample; Cohort -> Sample
+     *
+     * @param options           current query options object.
+     * @param projectionKeyList Projection keys to be removed from the query options.
+     * @return new QueryOptions after removing the inner projectionKey projections.
+     */
+    protected QueryOptions removeInnerProjections(QueryOptions options, List<String> projectionKeyList) {
         QueryOptions queryOptions = ParamUtils.defaultObject(options, QueryOptions::new);
 
         if (queryOptions.containsKey(QueryOptions.INCLUDE)) {
             List<String> includeList = queryOptions.getAsStringList(QueryOptions.INCLUDE);
             List<String> newInclude = new ArrayList<>(includeList.size());
-            boolean projectionKeyExcluded = false;
+            Map<String, Boolean> projectionKeyExcluded = new HashMap<>(projectionKeyList.size());
+            for (String projectionKey : projectionKeyList) {
+                projectionKeyExcluded.put(projectionKey, false);
+            }
+
             for (String include : includeList) {
-                if (!include.startsWith(projectionKey + ".")) {
+                boolean excluded = false;
+                for (String projectionKey : projectionKeyList) {
+                    if (include.startsWith(projectionKey + ".")) {
+                        projectionKeyExcluded.put(projectionKey, true);
+                        excluded = true;
+                        break;
+                    }
+                }
+                if (!excluded) {
                     newInclude.add(include);
-                } else {
-                    projectionKeyExcluded = true;
                 }
             }
             if (newInclude.isEmpty()) {
-                queryOptions.put(QueryOptions.INCLUDE, Arrays.asList(ID, projectionKey));
+                List<String> tmpInclude = new ArrayList<>(projectionKeyList.size() + 1);
+                tmpInclude.addAll(projectionKeyList);
+                tmpInclude.add(ID);
+                queryOptions.put(QueryOptions.INCLUDE, tmpInclude);
             } else {
-                if (projectionKeyExcluded) {
-                    newInclude.add(projectionKey);
+                for (Map.Entry<String, Boolean> entry : projectionKeyExcluded.entrySet()) {
+                    if (entry.getValue()) {
+                        newInclude.add(entry.getKey());
+                    }
                 }
                 queryOptions.put(QueryOptions.INCLUDE, newInclude);
             }
         }
-        if (queryOptions.containsKey(QueryOptions.EXCLUDE)) {
-            List<String> excludeList = queryOptions.getAsStringList(QueryOptions.EXCLUDE);
-            List<String> newExclude = new ArrayList<>(excludeList.size());
-            for (String exclude : excludeList) {
-                if (!exclude.startsWith(projectionKey + ".")) {
-                    newExclude.add(exclude);
-                }
-            }
-            if (newExclude.isEmpty()) {
-                queryOptions.remove(QueryOptions.EXCLUDE);
-            } else {
-                queryOptions.put(QueryOptions.EXCLUDE, newExclude);
-            }
-        }
+        // TODO: This code seems unnecessary. We should remove it if we don't find any issue.
+//        if (queryOptions.containsKey(QueryOptions.EXCLUDE)) {
+//            List<String> excludeList = queryOptions.getAsStringList(QueryOptions.EXCLUDE);
+//            List<String> newExclude = new ArrayList<>(excludeList.size());
+//            for (String exclude : excludeList) {
+//                if (!exclude.startsWith(projectionKey + ".")) {
+//                    newExclude.add(exclude);
+//                }
+//            }
+//            if (newExclude.isEmpty()) {
+//                queryOptions.remove(QueryOptions.EXCLUDE);
+//            } else {
+//                queryOptions.put(QueryOptions.EXCLUDE, newExclude);
+//            }
+//        }
 
         return queryOptions;
     }

opencga-catalog/src/main/java/org/opencb/opencga/catalog/managers/FamilyManager.java

@@ -83,6 +83,16 @@ public class FamilyManager extends AnnotationSetManager<Family> {
             FamilyDBAdaptor.QueryParams.ID.key(), FamilyDBAdaptor.QueryParams.UID.key(), FamilyDBAdaptor.QueryParams.UUID.key(),
             FamilyDBAdaptor.QueryParams.VERSION.key(), FamilyDBAdaptor.QueryParams.STUDY_UID.key(),
             FamilyDBAdaptor.QueryParams.MEMBERS.key()));
+    public static final QueryOptions INCLUDE_FAMILY_FOR_PEDIGREE = keepFieldsInQueryOptions(INCLUDE_FAMILY_IDS, Arrays.asList(
+            FamilyDBAdaptor.QueryParams.PEDIGREE_GRAPH.key(), FamilyDBAdaptor.QueryParams.DISORDERS.key(),
+            FamilyDBAdaptor.QueryParams.MEMBERS.key() + "." + IndividualDBAdaptor.QueryParams.ID.key(),
+            FamilyDBAdaptor.QueryParams.MEMBERS.key() + "." + IndividualDBAdaptor.QueryParams.SEX.key(),
+            FamilyDBAdaptor.QueryParams.MEMBERS.key() + "." + IndividualDBAdaptor.QueryParams.DISORDERS.key(),
+            FamilyDBAdaptor.QueryParams.MEMBERS.key() + "." + IndividualDBAdaptor.QueryParams.LIFE_STATUS.key(),
+            FamilyDBAdaptor.QueryParams.MEMBERS.key() + "." + IndividualDBAdaptor.QueryParams.FATHER.key()
+                    + "." + IndividualDBAdaptor.QueryParams.ID.key(),
+            FamilyDBAdaptor.QueryParams.MEMBERS.key() + "." + IndividualDBAdaptor.QueryParams.MOTHER.key()
+                    + "." + IndividualDBAdaptor.QueryParams.ID.key()));
     protected static Logger logger = LoggerFactory.getLogger(FamilyManager.class);
     private final String defaultFacet = "creationYear>>creationMonth;status;phenotypes;expectedSize;numMembers[0..20]:2";
     private UserManager userManager;