hotfix-4.5.x: new CLI parameter enables avoiding decomposition of MNV…

…s/block substitutions

cellbase-app/src/main/java/org/opencb/cellbase/app/cli/CliOptionsParser.java

@@ -383,13 +383,22 @@ public class VariantAnnotationCommandOptions {
                 required = false, arity = 1)
         public String referenceFasta;
 
-        @Parameter(names = {"--skip-normalize"}, description = "Skip normalization of input variants. Normalization"
-                + " includes splitting multi-allele positions read from a VCF, allele trimming and decomposing MNVs. Has"
+        @Parameter(names = {"--skip-normalize"}, description = "Skip normalization of input variants. Should not be used"
+                + " when the input (-i, --input-file) is a VCF file. Normalization includes splitting multi-allele positions "
+                + "read from a VCF, allele trimming and decomposing MNVs. Has"
                 + " no effect if reading variants from a CellBase variation collection "
                 + "(\"--input-variation-collection\") or running a variant annotation benchmark (\"--benchmark\"): in"
                 + " these two cases variant normalization is never carried out.",
                 required = false, arity = 0)
-        public boolean skipNormalize;
+        public boolean skipNormalize = false;
+
+        @Parameter(names = {"--skip-decompose"}, description = "Use this flag to avoid decomposition of "
+                + "multi-nucleotide-variants (MNVs) / block substitutions as part of the normalization process. If this"
+                + " flag is NOT activated, as a step during the normalization process reference and alternate alleles"
+                + " from MNVs/Block substitutions will be aligned and decomposed into their forming simple variants. "
+                + " This flag has no effect if --skip-normalize is present.",
+                required = false, arity = 0)
+        public boolean skipDecompose = false;
 
         @Parameter(names = {"--server-cache"}, description = "Use of this parameter is discouraged unless the "
                 + "server administrator advises so. Annotation was already pre-calculated and cached in "

cellbase-app/src/main/java/org/opencb/cellbase/app/cli/VariantAnnotationCommandExecutor.java

@@ -92,6 +92,7 @@ public enum FileFormat {VCF, JSON, AVRO, VEP};
     private boolean benchmark;
     private Path referenceFasta;
     private boolean normalize;
+    private boolean decompose;
     private List<String> chromosomeList;
     private int port;
     private String species;
@@ -379,14 +380,14 @@ private List<ParallelTaskRunner.TaskWithException<String, Variant, Exception>> g
                         VCFHeader header = (VCFHeader) codec.readActualHeader(lineIterator);
                         VCFHeaderVersion headerVersion = codec.getVCFHeaderVersion();
                         variantAnnotatorTaskList.add(new VcfStringAnnotatorTask(header, headerVersion,
-                                variantAnnotatorList, sharedContext, normalize));
+                                variantAnnotatorList, sharedContext, normalize, decompose));
                     } catch (IOException e) {
                         throw new IOException("Unable to read VCFHeader");
                     }
                     break;
                 case JSON:
                     logger.info("Using a JSON parser to read variants...");
-                    variantAnnotatorTaskList.add(new JsonStringAnnotatorTask(variantAnnotatorList, normalize));
+                    variantAnnotatorTaskList.add(new JsonStringAnnotatorTask(variantAnnotatorList, normalize, decompose));
                     break;
                 default:
                     break;
@@ -601,12 +602,17 @@ private void indexCustomVcfFile(int customFileNumber, RocksDB db) {
             Iterator<VariantContext> iterator = vcfFileReader.iterator();
             VariantContextToVariantConverter converter = new VariantContextToVariantConverter("", "",
                     vcfFileReader.getFileHeader().getSampleNamesInOrder());
-            VariantNormalizer normalizer = new VariantNormalizer(true, false, true);
+            // Currently, only VCF files are supported for custom-annotation so makes no sense to allow no normalisation
+            // of variants.
+            // However, decomposition of MNVs/Block substitutions can still be optional
+            VariantNormalizer normalizer = new VariantNormalizer(true, false, decompose);
             lineCounter = 0;
             while (iterator.hasNext()) {
                 variantContext = iterator.next();
                 // Reference positions will not be indexed
                 if (variantContext.getAlternateAlleles().size() > 0) {
+                    // Currently, only VCF files are supported for custom-annotation so makes no sense to allow no normalisation
+                    // of variants.
                     List<Variant> variantList = normalizer.normalize(converter.apply(Collections.singletonList(variantContext)), true);
                     for (Variant variant : variantList) {
                         db.put((variant.getChromosome() + "_" + variant.getStart() + "_" + variant.getReference() + "_"
@@ -740,6 +746,8 @@ private void checkParameters() throws IOException {
             normalize = false;
         }
 
+        decompose = !variantAnnotationCommandOptions.skipDecompose;
+
         // output file
         if (variantAnnotationCommandOptions.output != null) {
             output = Paths.get(variantAnnotationCommandOptions.output);

cellbase-app/src/main/java/org/opencb/cellbase/app/cli/variant/annotation/JsonStringAnnotatorTask.java

@@ -41,24 +41,25 @@ public class JsonStringAnnotatorTask implements ParallelTaskRunner.TaskWithExcep
     private List<VariantAnnotator> variantAnnotatorList;
     private boolean normalize;
     private static ObjectMapper jsonObjectMapper;
-    private static VariantNormalizer normalizer;
+    private VariantNormalizer normalizer;
 
     static {
         jsonObjectMapper = new ObjectMapper();
         jsonObjectMapper.configure(MapperFeature.REQUIRE_SETTERS_FOR_GETTERS, true);
         jsonObjectMapper.setSerializationInclusion(JsonInclude.Include.NON_NULL);
-        normalizer = new VariantNormalizer(true, false, true);
     }
 
     private static final String VARIANT_STRING_PATTERN = "([ACGTN]*)|(<CNV[0-9]+>)|(<DUP>)|(<DEL>)|(<INS>)|(<INV>)";
 
     public JsonStringAnnotatorTask(List<VariantAnnotator> variantAnnotatorList) {
-        this(variantAnnotatorList, true);
+        this(variantAnnotatorList, true, true);
     }
 
-    public JsonStringAnnotatorTask(List<VariantAnnotator> variantAnnotatorList, boolean normalize) {
+    public JsonStringAnnotatorTask(List<VariantAnnotator> variantAnnotatorList, boolean normalize,
+                                   boolean decompose) {
         this.variantAnnotatorList = variantAnnotatorList;
         this.normalize = normalize;
+        normalizer = new VariantNormalizer(true, false, decompose);
     }
 
     public void pre() {

cellbase-app/src/main/java/org/opencb/cellbase/app/cli/variant/annotation/VcfStringAnnotatorTask.java

@@ -49,23 +49,25 @@ public class VcfStringAnnotatorTask implements ParallelTaskRunner.TaskWithExcept
     private List<VariantAnnotator> variantAnnotatorList;
     private FullVcfCodec vcfCodec;
     private VariantContextToVariantConverter converter;
-    private static VariantNormalizer normalizer = new VariantNormalizer(true, false, true);
+//    private static VariantNormalizer normalizer = new VariantNormalizer(true, false, true);
+    private VariantNormalizer normalizer;
     private boolean normalize;
 
     public VcfStringAnnotatorTask(VCFHeader header, VCFHeaderVersion version,
                                   List<VariantAnnotator> variantAnnotatorList, SharedContext sharedContext) {
-        this(header, version, variantAnnotatorList, sharedContext, true);
+        this(header, version, variantAnnotatorList, sharedContext, true, true);
     }
 
     public VcfStringAnnotatorTask(VCFHeader header, VCFHeaderVersion version,
                                   List<VariantAnnotator> variantAnnotatorList, SharedContext sharedContext,
-                                  boolean normalize) {
+                                  boolean normalize, boolean decompose) {
         this.vcfCodec = new FullVcfCodec();
         this.vcfCodec.setVCFHeader(header, version);
         this.converter = new VariantContextToVariantConverter("", "", header.getSampleNamesInOrder());
         this.variantAnnotatorList = variantAnnotatorList;
         this.sharedContext = sharedContext;
         this.normalize = normalize;
+        normalizer = new VariantNormalizer(true, false, decompose);
     }
 
     @Override