Merge pull request #41 from molgenis/feat/more_possible

Variants without a gene and pedigrees with missing GT should have possible match

src/main/java/org/molgenis/vcf/inheritance/matcher/Annotator.java

@@ -1,6 +1,7 @@
 package org.molgenis.vcf.inheritance.matcher;
 
 import static org.molgenis.vcf.inheritance.matcher.model.InheritanceMatch.TRUE;
+import static org.molgenis.vcf.inheritance.matcher.model.InheritanceMatch.POTENTIAL;
 import static org.molgenis.vcf.utils.utils.HeaderUtils.fixVcfFilterHeaderLines;
 import static org.molgenis.vcf.utils.utils.HeaderUtils.fixVcfFormatHeaderLines;
 import static org.molgenis.vcf.utils.utils.HeaderUtils.fixVcfInfoHeaderLines;
@@ -57,7 +58,7 @@ VCFHeader annotateHeader(VCFHeader vcfHeader) {
         "Inheritance Match: Genotypes, affected statuses and known gene inheritance patterns match."));
     vcfHeader.addMetaDataLine(new VCFFormatHeaderLine(MATCHING_GENES, VCFHeaderLineCount.UNBOUNDED,
         VCFHeaderLineType.String,
-        "Genes with an inheritance match."));
+        "Genes with a (potential) inheritance match."));
 
     Set<VCFHeaderLine> headerLines = new LinkedHashSet<>();
     //workaround for "Escaped doublequotes in INFO descriptions result in invalid VCF file"
@@ -112,7 +113,7 @@ private void annotateGenotype(VariantContext vc, Annotation annotation,
       String inheritanceMatch = mapInheritanceMatch(match);
       genotypeBuilder
               .attribute(INHERITANCE_MATCH, inheritanceMatch);
-      if (match == TRUE) {
+      if ((match == TRUE || match == POTENTIAL) && annotation.getMatchingGenes() != null && !annotation.getMatchingGenes().isEmpty()) {
         genotypeBuilder
             .attribute(MATCHING_GENES, annotation.getMatchingGenes().stream().sorted().collect(
                 Collectors.joining(",")));
@@ -127,7 +128,7 @@ private static String mapInheritanceMatch(InheritanceMatch match) {
     switch (match){
       case TRUE -> inheritanceMatch = "1";
       case FALSE -> inheritanceMatch = "0";
-      case UNKNOWN -> inheritanceMatch = null;
+      case POTENTIAL -> inheritanceMatch = null;
       default -> throw new UnexpectedEnumException(match);
     }
     return inheritanceMatch;

src/main/java/org/molgenis/vcf/inheritance/matcher/InheritanceMatcher.java

@@ -1,6 +1,5 @@
 package org.molgenis.vcf.inheritance.matcher;
 
-import java.util.Collection;
 import java.util.HashMap;
 import java.util.HashSet;
 import java.util.Map;
@@ -17,7 +16,7 @@ private InheritanceMatcher() {
   }
 
   public static Map<String, Annotation> matchInheritance(
-      Map<String, Inheritance> inheritanceMap, Collection<Gene> genes) {
+      Map<String, Inheritance> inheritanceMap, VariantContextGenes genes) {
     Map<String, Annotation> sampleAnnotationMap = new HashMap<>();
     for (Entry<String, Inheritance> entry : inheritanceMap.entrySet()) {
       Set<String> matchingGenes = new HashSet<>();
@@ -44,9 +43,9 @@ public static Map<String, Annotation> matchInheritance(
      *  - inheritance match is unknown if any genes for the variant have unknown inheritance pattern.
      *  - inheritance match is false if all genes for the variant have known (but mismatching) inheritance pattern.
      */
-    private static void matchGeneInheritance(Collection<Gene> genes, Set<String> matchingGenes, Inheritance inheritance) {
+    private static void matchGeneInheritance(VariantContextGenes genes, Set<String> matchingGenes, Inheritance inheritance) {
         boolean containsUnknownGene = false;
-        for (Gene gene : genes) {
+        for (Gene gene : genes.getGenes().values()) {
           Set<InheritanceMode> geneInheritanceModes = gene
                   .getInheritanceModes();
           if( geneInheritanceModes.isEmpty() ){
@@ -55,12 +54,16 @@ private static void matchGeneInheritance(Collection<Gene> genes, Set<String> mat
           if (geneInheritanceModes.stream()
                       .anyMatch(mode -> inheritance.getInheritanceModes().contains(mode))) {
               matchingGenes.add(gene.getId());
-              inheritance.setMatch(TRUE);
+              if(inheritance.isFamilyWithMissingGT()){
+                  inheritance.setMatch(POTENTIAL);
+              }else {
+                  inheritance.setMatch(TRUE);
+              }
           }
         }
         if(matchingGenes.isEmpty()) {
-            if (containsUnknownGene) {
-                inheritance.setMatch(UNKNOWN);
+            if (containsUnknownGene || genes.isContainsVcWithoutGene()) {
+                inheritance.setMatch(POTENTIAL);
             } else {
                 inheritance.setMatch(FALSE);
             }

src/main/java/org/molgenis/vcf/inheritance/matcher/InheritanceService.java

@@ -6,6 +6,7 @@
 import static org.molgenis.vcf.utils.sample.mapper.PedToSamplesMapper.mapPedFileToPedigrees;
 
 import htsjdk.variant.variantcontext.Allele;
+import htsjdk.variant.variantcontext.Genotype;
 import htsjdk.variant.variantcontext.VariantContext;
 import htsjdk.variant.variantcontext.writer.VariantContextWriter;
 import htsjdk.variant.variantcontext.writer.VariantContextWriterBuilder;
@@ -27,12 +28,7 @@
 import org.molgenis.vcf.inheritance.matcher.checker.DeNovoChecker;
 import org.molgenis.vcf.inheritance.matcher.checker.XldChecker;
 import org.molgenis.vcf.inheritance.matcher.checker.XlrChecker;
-import org.molgenis.vcf.inheritance.matcher.model.Annotation;
-import org.molgenis.vcf.inheritance.matcher.model.Gene;
-import org.molgenis.vcf.inheritance.matcher.model.Inheritance;
-import org.molgenis.vcf.inheritance.matcher.model.InheritanceMode;
-import org.molgenis.vcf.inheritance.matcher.model.Settings;
-import org.molgenis.vcf.inheritance.matcher.model.SubInheritanceMode;
+import org.molgenis.vcf.inheritance.matcher.model.*;
 import org.molgenis.vcf.inheritance.matcher.util.InheritanceUtils;
 import org.molgenis.vcf.utils.metadata.FieldMetadataService;
 import org.molgenis.vcf.utils.sample.model.AffectedStatus;
@@ -103,7 +99,7 @@ private VariantContext processSingleVariantcontext(List<String> probands, VepMap
     Map<String, Inheritance> inheritanceMap = matchInheritanceForVariant(geneVariantMap,
         vc, pedigreeList, probands);
     Map<String, Annotation> annotationMap = matchInheritance(inheritanceMap,
-        vepMapper.getGenes(vc).values());
+        vepMapper.getGenes(vc));
     return annotator.annotateInheritance(vc, pedigreeList, annotationMap);
   }
 
@@ -112,9 +108,9 @@ private Map<String, List<VariantContext>> createGeneVariantMap(VepMapper vepMapp
       List<VariantContext> variantContextList) {
     Map<String, List<VariantContext>> geneVariantMap = new HashMap<>();
     for (VariantContext vc : variantContextList) {
-      Map<String, Gene> genes = vepMapper.getGenes(vc, knownGenes);
-      knownGenes.putAll(genes);
-      for (Gene gene : genes.values()) {
+      VariantContextGenes variantContextGenes = vepMapper.getGenes(vc, knownGenes);
+      knownGenes.putAll(variantContextGenes.getGenes());
+      for (Gene gene : variantContextGenes.getGenes().values()) {
         List<VariantContext> geneVariantList;
         if (geneVariantMap.containsKey(gene.getId())) {
           geneVariantList = geneVariantMap.get(gene.getId());
@@ -172,10 +168,18 @@ private Inheritance matchInheritanceForSample(
     checkAd(variantContext, filteredFamily, inheritance);
     checkXl(variantContext, filteredFamily, inheritance);
     inheritance.setDenovo(deNovoChecker.checkDeNovo(variantContext, filteredFamily, sample));
+    inheritance.setFamilyWithMissingGT(checkMissingGTs(filteredFamily, variantContext));
 
     return inheritance;
   }
 
+  private boolean checkMissingGTs(Pedigree filteredFamily,VariantContext variantContext) {
+    return filteredFamily.getMembers().values().stream().anyMatch(sample -> {
+      Genotype gt = variantContext.getGenotype(sample.getPerson().getIndividualId());
+      return gt == null || !gt.isCalled();
+    });
+  }
+
   private void checkXl(VariantContext variantContext, Pedigree family,
       Inheritance inheritance) {
     if (xldChecker.check(variantContext, family)) {

src/main/java/org/molgenis/vcf/inheritance/matcher/VepMapper.java

@@ -1,6 +1,5 @@
 package org.molgenis.vcf.inheritance.matcher;
 
-import static java.util.Arrays.asList;
 import static java.util.Collections.emptyMap;
 
 import htsjdk.variant.variantcontext.VariantContext;
@@ -14,6 +13,7 @@
 import java.util.Set;
 import org.molgenis.vcf.inheritance.matcher.model.Gene;
 import org.molgenis.vcf.inheritance.matcher.model.InheritanceMode;
+import org.molgenis.vcf.inheritance.matcher.model.VariantContextGenes;
 import org.molgenis.vcf.utils.metadata.FieldMetadataService;
 import org.molgenis.vcf.utils.model.FieldMetadata;
 import org.molgenis.vcf.utils.model.NestedField;
@@ -63,18 +63,20 @@ private void init(VCFFileReader vcfFileReader) {
     throw new MissingInfoException("VEP");
   }
 
-  public Map<String, Gene> getGenes(VariantContext vc) {
+  public VariantContextGenes getGenes(VariantContext vc) {
     return getGenes(vc, emptyMap());
   }
 
-  public Map<String, Gene> getGenes(VariantContext vc, Map<String, Gene> knownGenes) {
+  public VariantContextGenes getGenes(VariantContext vc, Map<String, Gene> knownGenes) {
+    VariantContextGenes.VariantContextGenesBuilder genesBuilder = VariantContextGenes.builder();
     Map<String, Gene> genes = new HashMap<>();
     List<String> vepValues = vc.getAttributeAsStringList(vepFieldId, "");
     for (String vepValue : vepValues) {
       String[] vepSplit = vepValue.split("\\|", -1);
       String gene = vepSplit[geneIndex];
       String source = vepSplit[geneSourceIndex];
       if (gene.isEmpty() || source.isEmpty()) {
+        genesBuilder.containsVcWithoutGene(true);
         continue;
       }
 
@@ -94,7 +96,8 @@ public Map<String, Gene> getGenes(VariantContext vc, Map<String, Gene> knownGene
         genes.put(gene, knownGenes.get(gene));
       }
     }
-    return genes;
+    genesBuilder.genes(genes);
+    return genesBuilder.build();
   }
 
   private void mapGeneInheritance(Set<InheritanceMode> modes, String[] inheritanceModes) {
@@ -123,7 +126,7 @@ private void mapGeneInheritance(Set<InheritanceMode> modes, String[] inheritance
   }
 
   public boolean containsIncompletePenetrance(VariantContext variantContext) {
-    Map<String, Gene> genes = getGenes(variantContext);
+    Map<String, Gene> genes = getGenes(variantContext).getGenes();
     return genes.values().stream().anyMatch(Gene::isIncompletePenetrance);
   }
 }

src/main/java/org/molgenis/vcf/inheritance/matcher/checker/ArCompoundChecker.java

@@ -27,7 +27,7 @@ public List<VariantContext> check(
       VariantContext variantContext, Pedigree family) {
     if (onAutosome(variantContext)) {
       List<VariantContext> compounds = new ArrayList<>();
-      Map<String, Gene> genes = vepMapper.getGenes(variantContext);
+      Map<String, Gene> genes = vepMapper.getGenes(variantContext).getGenes();
       for (Gene gene : genes.values()) {
         checkForGene(geneVariantMap, variantContext, family, compounds, gene);
       }

src/main/java/org/molgenis/vcf/inheritance/matcher/model/Inheritance.java

@@ -16,11 +16,14 @@ public class Inheritance {
   Set<String> compounds = new HashSet<>();
 
   @Builder.Default
-  InheritanceMatch match = InheritanceMatch.UNKNOWN;
+  InheritanceMatch match = InheritanceMatch.POTENTIAL;
 
   @Builder.Default
   boolean denovo = false;
 
+  @Builder.Default
+  boolean isFamilyWithMissingGT = false;
+
   public void addInheritanceMode(InheritanceMode inheritanceMode) {
     inheritanceModes.add(inheritanceMode);
   }

src/main/java/org/molgenis/vcf/inheritance/matcher/model/InheritanceMatch.java

@@ -1,5 +1,5 @@
 package org.molgenis.vcf.inheritance.matcher.model;
 
 public enum InheritanceMatch {
-    TRUE, FALSE, UNKNOWN
+    TRUE, FALSE, POTENTIAL
 }

src/main/java/org/molgenis/vcf/inheritance/matcher/model/VariantContextGenes.java

@@ -0,0 +1,12 @@
+package org.molgenis.vcf.inheritance.matcher.model;
+
+import lombok.Builder;
+import lombok.Data;
+import java.util.Map;
+@Data
+@Builder
+public class VariantContextGenes{
+    Map<String, Gene> genes;
+    @Builder.Default
+    boolean containsVcWithoutGene = false;
+}

src/test/java/org/molgenis/vcf/inheritance/matcher/AnnotatorTest.java

@@ -79,7 +79,7 @@ void annotateHeader() {
         () -> verify(vcfHeader)
             .addMetaDataLine(new VCFFormatHeaderLine(MATCHING_GENES, VCFHeaderLineCount.UNBOUNDED,
                 VCFHeaderLineType.String,
-                "Genes with an inheritance match.")),
+                "Genes with a (potential) inheritance match.")),
         () -> verify(vcfHeader, times(2)).getInfoHeaderLines(),
         () -> verify(vcfHeader, times(2)).getFormatHeaderLines(),
         () -> verify(vcfHeader, times(2)).getFilterLines(),

src/test/java/org/molgenis/vcf/inheritance/matcher/InheritanceMatcherTest.java

@@ -7,14 +7,10 @@
 import static org.molgenis.vcf.inheritance.matcher.model.InheritanceMode.AR;
 import static org.molgenis.vcf.inheritance.matcher.model.InheritanceMode.AD;
 
-import java.util.Collection;
 import java.util.Map;
 import java.util.Set;
 import org.junit.jupiter.api.Test;
-import org.molgenis.vcf.inheritance.matcher.model.Annotation;
-import org.molgenis.vcf.inheritance.matcher.model.Gene;
-import org.molgenis.vcf.inheritance.matcher.model.Inheritance;
-import org.molgenis.vcf.inheritance.matcher.model.SubInheritanceMode;
+import org.molgenis.vcf.inheritance.matcher.model.*;
 
 class InheritanceMatcherTest {
 
@@ -26,7 +22,7 @@ void matchInheritanceMatch() {
         Set.of(AR)).subInheritanceModes(emptySet()).compounds(singleton("OTHER_VARIANT")).build();
     Map<String, Inheritance> inheritanceMap = Map.of("sample1", inheritance1, "sample2",
         inheritance2);
-    Collection<Gene> genes = Set.of(new Gene("GENE1","EntrezGene", true, Set.of(AR,AD)),new Gene("GENE2","EntrezGene", true, Set.of(AD)));
+    VariantContextGenes genes = VariantContextGenes.builder().genes(Map.of("GENE1", new Gene("GENE1","EntrezGene", true, Set.of(AR,AD)), "GENE2", new Gene("GENE2","EntrezGene", true, Set.of(AD)))).build();
 
     Map<String, Annotation> actual = InheritanceMatcher
         .matchInheritance(inheritanceMap, genes);
@@ -46,7 +42,7 @@ void matchInheritanceMismatch() {
     Inheritance inheritance = Inheritance.builder().inheritanceModes(
             Set.of(AD)).subInheritanceModes(Set.of(SubInheritanceMode.AD_IP)).build();
     Map<String, Inheritance> inheritanceMap = Map.of("sample1", inheritance);
-    Collection<Gene> genes = Set.of(new Gene("GENE1","EntrezGene", true, Set.of(AR)));
+    VariantContextGenes genes = VariantContextGenes.builder().genes(Map.of("GENE1", new Gene("GENE1","EntrezGene", true, Set.of(AR)))).build();
 
     Map<String, Annotation> actual = InheritanceMatcher
             .matchInheritance(inheritanceMap, genes);
@@ -64,7 +60,7 @@ void matchInheritanceNoSuitableModes() {
     Inheritance inheritance1 = Inheritance.builder().match(FALSE).inheritanceModes(
             emptySet()).subInheritanceModes(emptySet()).build();
     Map<String, Inheritance> inheritanceMap = Map.of("sample1", inheritance1);
-    Collection<Gene> genes = Set.of(new Gene("GENE1","EntrezGene", false, Set.of(AR,AD)),new Gene("GENE2","EntrezGene", false, Set.of(AD)));
+    VariantContextGenes genes = VariantContextGenes.builder().genes(Map.of("GENE1",new Gene("GENE1","EntrezGene", false, Set.of(AR,AD)),"GENE2",new Gene("GENE2","EntrezGene", false, Set.of(AD)))).build();
 
     Map<String, Annotation> actual = InheritanceMatcher
             .matchInheritance(inheritanceMap, genes);
@@ -81,12 +77,12 @@ void matchInheritanceUnknownGene() {
     Inheritance inheritance1 = Inheritance.builder().inheritanceModes(
             Set.of(AD,AR)).subInheritanceModes(Set.of(SubInheritanceMode.AD_IP, SubInheritanceMode.AR_C)).compounds(singleton("OTHER_VARIANT")).build();
     Map<String, Inheritance> inheritanceMap = Map.of("sample1", inheritance1);
-    Collection<Gene> genes = Set.of(new Gene("GENE1","EntrezGene", true, emptySet()));
+    VariantContextGenes genes = VariantContextGenes.builder().genes(Map.of("GENE1",new Gene("GENE1","EntrezGene", true, emptySet()))).build();
 
     Map<String, Annotation> actual = InheritanceMatcher
             .matchInheritance(inheritanceMap, genes);
 
-    Inheritance expectedInheritance = Inheritance.builder().match(UNKNOWN).inheritanceModes(
+    Inheritance expectedInheritance = Inheritance.builder().match(POTENTIAL).inheritanceModes(
             Set.of(AR, AD)).subInheritanceModes(Set.of(SubInheritanceMode.AD_IP, SubInheritanceMode.AR_C)).compounds(singleton("OTHER_VARIANT")).build();
     Annotation expectedAnnotation = Annotation.builder().inheritance(expectedInheritance).matchingGenes(emptySet()).build();
     Map<String, Annotation> expected = Map.of("sample1",expectedAnnotation);
@@ -98,7 +94,7 @@ void matchInheritanceUnknownGeneAndKnownGeneMatch() {
     Inheritance inheritance1 = Inheritance.builder().inheritanceModes(
             Set.of(AR)).subInheritanceModes(Set.of(SubInheritanceMode.AR_C)).compounds(singleton("OTHER_VARIANT")).build();
     Map<String, Inheritance> inheritanceMap = Map.of("sample1", inheritance1);
-    Collection<Gene> genes = Set.of(new Gene("GENE1","EntrezGene", false, emptySet()),new Gene("GENE2","EntrezGene", false, Set.of(AR)));
+    VariantContextGenes genes = VariantContextGenes.builder().genes(Map.of("GENE1",new Gene("GENE1","EntrezGene", false, emptySet()),"GENE2",new Gene("GENE2","EntrezGene", false, Set.of(AR)))).build();
 
     Map<String, Annotation> actual = InheritanceMatcher
             .matchInheritance(inheritanceMap, genes);
@@ -116,12 +112,12 @@ void matchInheritanceUnknownGeneAndKnownGeneMismatch() {
             Set.of(AD)).subInheritanceModes(emptySet()).compounds(emptySet()).build();
 
     Map<String, Inheritance> inheritanceMap = Map.of("sample1", inheritance1);
-    Collection<Gene> genes = Set.of(new Gene("GENE1","EntrezGene", false, emptySet()),new Gene("GENE2","EntrezGene", false, Set.of(AR)));
+    VariantContextGenes genes = VariantContextGenes.builder().genes(Map.of("GENE1", new Gene("GENE1","EntrezGene", false, emptySet()),"GENE2", new Gene("GENE2","EntrezGene", false, Set.of(AR)))).build();
 
     Map<String, Annotation> actual = InheritanceMatcher
             .matchInheritance(inheritanceMap, genes);
 
-    Inheritance expectedInheritance = Inheritance.builder().match(UNKNOWN).inheritanceModes(
+    Inheritance expectedInheritance = Inheritance.builder().match(POTENTIAL).inheritanceModes(
             Set.of(AD)).subInheritanceModes(emptySet()).compounds(emptySet()).build();
     Annotation expectedAnnotation = Annotation.builder().inheritance(expectedInheritance).matchingGenes(emptySet()).build();
     Map<String, Annotation> expected = Map.of("sample1",expectedAnnotation);