SCRUM-5881 use severity rank for sorting molecular consequences (#1550)

Author: cmpich

agr_java_core/src/main/java/org/alliancegenome/core/variant/converters/VariantSummaryConverter.java

@@ -3,6 +3,7 @@
 import java.net.URLDecoder;
 import java.nio.charset.StandardCharsets;
 import java.util.ArrayList;
+import java.util.Comparator;
 import java.util.HashSet;
 import java.util.List;
 import java.util.Map;
@@ -45,6 +46,7 @@ public class VariantSummaryConverter {
 	// Header index positions (initialized once per header)
 	private String[] header;
 	private GeneDocumentCache geneCache;
+	private Map<String, Integer> severityRanking;
 	private Map<String, SOTerm> soTermCache = new ConcurrentHashMap<>();
 	private Map<String, VocabularyTerm> vocabularyTermCache = new ConcurrentHashMap<>();
 
@@ -71,9 +73,10 @@ public class VariantSummaryConverter {
 	private int proteinPosIdx = -1;
 	private int hgvsgIdx = -1;
 
-	public VariantSummaryConverter(String[] header, GeneDocumentCache geneCache) {
+	public VariantSummaryConverter(String[] header, GeneDocumentCache geneCache, Map<String, Integer> severityRanking) {
 		this.header = header;
 		this.geneCache = geneCache;
+		this.severityRanking = severityRanking;
 		initializeHeaderIndices();
 	}
 
@@ -347,6 +350,9 @@ private Pair<List<PredictedVariantConsequence>, HashSet<String>> getConsequences
 						ampIdx = csqField.indexOf('&', start);
 					} while (ampIdx >= 0);
 					soTerms.add(getSOTerm(csqField.substring(start)));
+					soTerms.sort(Comparator.comparingInt(
+						term -> term.getSeverityOrder() != null ? term.getSeverityOrder() : Integer.MAX_VALUE
+					));
 					consequence.setVepConsequences(soTerms);
 				}
 			}
@@ -559,6 +565,9 @@ private SOTerm getSOTerm(String name) {
 		return soTermCache.computeIfAbsent(name, k -> {
 			SOTerm term = new SOTerm();
 			term.setName(k);
+			if (severityRanking != null) {
+				term.setSeverityOrder(severityRanking.get(k));
+			}
 			return term;
 		});
 	}

agr_variant_indexer/src/main/java/org/alliancegenome/indexer/variant/es/managers/SourceDocumentCreation.java

@@ -4,6 +4,7 @@
 import java.util.ArrayList;
 import java.util.HashSet;
 import java.util.List;
+import java.util.Map;
 import java.util.concurrent.LinkedBlockingDeque;
 import java.util.concurrent.TimeUnit;
 
@@ -38,6 +39,7 @@ public class SourceDocumentCreation extends Thread {
 
 	private final GeneDocumentCache geneCache;
 	private final HashSet<String> variantsCache;
+	private final Map<String, Integer> severityRanking;
 	private String downloadPath;
 	private DownloadSource source;
 	private SpeciesType speciesType;
@@ -60,11 +62,12 @@ public class SourceDocumentCreation extends Thread {
 
 	private String messageHeader = "";
 
-	public SourceDocumentCreation(String downloadPath, DownloadSource source, GeneDocumentCache geneCache, HashSet<String> variantsCache) {
+	public SourceDocumentCreation(String downloadPath, DownloadSource source, GeneDocumentCache geneCache, HashSet<String> variantsCache, Map<String, Integer> severityRanking) {
 		this.downloadPath = downloadPath;
 		this.source = source;
 		this.geneCache = geneCache;
 		this.variantsCache = variantsCache;
+		this.severityRanking = severityRanking;
 		speciesType = SpeciesType.getTypeByID(source.getTaxonId());
 		messageHeader = speciesType.getModName() + " ";
 		int vcQueueSize = source.getVcQueueSize() != null ? source.getVcQueueSize() : VariantConfigHelper.getSourceDocumentCreatorVCQueueSize();
@@ -198,7 +201,7 @@ public void run() {
 				VCFInfoHeaderLine fileHeader = reader.getFileHeader().getInfoHeaderLine("CSQ");
 				header = fileHeader.getDescription().split("Format: ")[1].split("\\|");
 				// All files for a Mod have the same header so we only need one of them
-				variantSummaryConverter = new VariantSummaryConverter(header, geneCache);
+				variantSummaryConverter = new VariantSummaryConverter(header, geneCache, severityRanking);
 				sequenceSummaryConverter = new SequenceSummaryConverter();
 				variantSearchResultConverter = new VariantSearchResultConverter();
 				try {

agr_variant_indexer/src/main/java/org/alliancegenome/indexer/variant/es/managers/SourceDocumentCreationManager.java

@@ -4,10 +4,12 @@
 import java.util.ArrayList;
 import java.util.HashSet;
 import java.util.List;
+import java.util.Map;
 
 import org.alliancegenome.core.config.ConfigHelper;
 import org.alliancegenome.core.filedownload.model.DownloadFileSet;
 import org.alliancegenome.core.filedownload.model.DownloadSource;
+import org.alliancegenome.curation_api.interfaces.crud.ontology.SoTermCrudInterface;
 import org.alliancegenome.curation_api.interfaces.document.VariantDocumentInterface;
 import org.alliancegenome.es.index.site.cache.GeneDocumentCache;
 import org.alliancegenome.es.rest.RestConfig;
@@ -24,6 +26,7 @@ public class SourceDocumentCreationManager extends Thread {
 	private DownloadFileSet downloadSet;
 
 	private final VariantDocumentInterface variantApi = RestProxyFactory.createProxy(VariantDocumentInterface.class, ConfigHelper.getCurationApiUrl(), RestConfig.config);
+	private final SoTermCrudInterface soTermApi = RestProxyFactory.createProxy(SoTermCrudInterface.class, ConfigHelper.getCurationApiUrl(), RestConfig.config);
 
 	public SourceDocumentCreationManager(DownloadFileSet downloadSet) {
 		this.downloadSet = downloadSet;
@@ -56,10 +59,13 @@ public void run() {
 				e.printStackTrace();
 			}
 
+			log.info("Fetching SO term severity ranking from curation API...");
+			Map<String, Integer> severityRanking = soTermApi.getSeverityRanking();
+			log.info("Fetched severity ranking for {} SO terms", severityRanking.size());
 			List<SourceDocumentCreation> creators = new ArrayList<>();
 			for (DownloadSource source : downloadSet.getDownloadFileSources()) {
 				if (source.getActive()) {
-					SourceDocumentCreation creator = new SourceDocumentCreation(downloadSet.getDownloadPath(), source, geneCache, variantsCache);
+					SourceDocumentCreation creator = new SourceDocumentCreation(downloadSet.getDownloadPath(), source, geneCache, variantsCache, severityRanking);
 					creator.start();
 					creators.add(creator);
 				}