Merge pull request #101 from molgenis/feat/utils3.0.0

Move to vip utils 3.0.0
molgenis · Nov 26, 2024 · 28f9034 · 28f9034
2 parents 277285c + 491a14d
commit 28f9034
Show file tree

Hide file tree

Showing 45 changed files with 535 additions and 612 deletions.
diff --git a/pom.xml b/pom.xml
@@ -43,7 +43,7 @@
         <commons.cli.version>1.9.0</commons.cli.version>
         <samtools.htsjdk.version>4.1.1</samtools.htsjdk.version>
         <jacoco-maven-plugin.version>0.8.12</jacoco-maven-plugin.version>
-        <vip.utils.version>2.0.2</vip.utils.version>
+        <vip.utils.version>3.0.0</vip.utils.version>
     </properties>
 
     <profiles>

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/VcfMetadata.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/VcfMetadata.java
@@ -18,12 +18,11 @@
 import org.molgenis.vcf.decisiontree.filter.model.FieldImpl;
 import org.molgenis.vcf.decisiontree.filter.model.FieldType;
 import org.molgenis.vcf.decisiontree.filter.model.MissingField;
-import org.molgenis.vcf.decisiontree.filter.model.ValueCount;
-import org.molgenis.vcf.decisiontree.filter.model.ValueCount.Type;
-import org.molgenis.vcf.decisiontree.filter.model.ValueCount.ValueCountBuilder;
-import org.molgenis.vcf.decisiontree.filter.model.ValueType;
 import org.molgenis.vcf.decisiontree.runner.info.NestedHeaderLine;
 import org.molgenis.vcf.utils.UnexpectedEnumException;
+import org.molgenis.vcf.utils.metadata.ValueCount;
+import org.molgenis.vcf.utils.metadata.ValueCount.ValueCountBuilder;
+import org.molgenis.vcf.utils.metadata.ValueType;
 
 /**
  * {@link VCFHeader} wrapper that works with nested metadata (e.g. CSQ INFO fields).
@@ -72,15 +71,15 @@ private static Field toSampleField(List<String> fieldTokens) {
     switch (field.toUpperCase()) {
       case "PROBAND" -> {
         valueType = ValueType.FLAG;
-        valueCount = ValueCount.builder().type(Type.FIXED).count(1).build();
+        valueCount = ValueCount.builder().type(ValueCount.Type.FIXED).count(1).build();
       }
       case "ID", "AFFECTED_STATUS", "SEX", "FATHER_ID", "MOTHER_ID", "FAMILY_ID" -> {
         valueType = ValueType.STRING;
-        valueCount = ValueCount.builder().type(Type.FIXED).count(1).build();
+        valueCount = ValueCount.builder().type(ValueCount.Type.FIXED).count(1).build();
       }
       case "PHENOTYPES" -> {
         valueType = ValueType.STRING;
-        valueCount = ValueCount.builder().type(Type.VARIABLE).nullable(true).build();
+        valueCount = ValueCount.builder().type(ValueCount.Type.VARIABLE).nullable(true).build();
       }
       default -> throw new UnsupportedFieldException(field);
     }
@@ -127,19 +126,19 @@ private FieldImpl toCommonField(List<String> fieldTokens) {
     switch (field) {
       case "#CHROM", "REF" -> {
         valueType = ValueType.STRING;
-        valueCount = ValueCount.builder().type(Type.FIXED).count(1).build();
+        valueCount = ValueCount.builder().type(ValueCount.Type.FIXED).count(1).build();
       }
       case "POS" -> {
         valueType = ValueType.INTEGER;
-        valueCount = ValueCount.builder().type(Type.FIXED).count(1).build();
+        valueCount = ValueCount.builder().type(ValueCount.Type.FIXED).count(1).build();
       }
       case "ID", "FILTER", "ALT" -> {
         valueType = ValueType.STRING;
-        valueCount = ValueCount.builder().type(Type.VARIABLE).nullable(true).build();
+        valueCount = ValueCount.builder().type(ValueCount.Type.VARIABLE).nullable(true).build();
       }
       case "QUAL" -> {
         valueType = ValueType.FLOAT;
-        valueCount = ValueCount.builder().type(Type.FIXED).count(1).nullable(true).build();
+        valueCount = ValueCount.builder().type(ValueCount.Type.FIXED).count(1).nullable(true).build();
       }
       default -> throw new UnsupportedFieldException(field);
     }
@@ -179,12 +178,12 @@ private Field toCompoundField(List<String> fieldTokens, FieldType fieldType) {
     switch (countType) {
       case INTEGER -> {
         int count = vcfCompoundHeaderLine.getCount();
-        builder.type(Type.FIXED).count(count).nullable(valueType != ValueType.FLAG);
+        builder.type(ValueCount.Type.FIXED).count(count).nullable(valueType != ValueType.FLAG);
       }
-      case A -> builder.type(Type.A).nullable(true);
-      case R -> builder.type(Type.R).nullable(true);
-      case G -> builder.type(Type.G).nullable(true);
-      case UNBOUNDED -> builder.type(Type.VARIABLE).nullable(true);
+      case A -> builder.type(ValueCount.Type.A).nullable(true);
+      case R -> builder.type(ValueCount.Type.R).nullable(true);
+      case G -> builder.type(ValueCount.Type.G).nullable(true);
+      case UNBOUNDED -> builder.type(ValueCount.Type.VARIABLE).nullable(true);
       default -> throw new UnexpectedEnumException(countType);
     }
 
@@ -217,7 +216,7 @@ private VCFCompoundHeaderLine getVcfCompoundHeaderLine(FieldType fieldType, Stri
   }
 
   public static boolean isSingleValueField(Field field) {
-    return field.getValueCount().getType() == Type.FIXED && field.getValueCount().getCount() <= 1;
+    return field.getValueCount().getType() == ValueCount.Type.FIXED && field.getValueCount().getCount() <= 1;
   }
 
   public Map<String, Integer> getSampleNameToOffset() {

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/VcfReader.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/VcfReader.java
@@ -5,40 +5,56 @@
 import htsjdk.variant.vcf.VCFFileReader;
 import java.util.stream.Stream;
 import java.util.stream.StreamSupport;
+
+import htsjdk.variant.vcf.VCFInfoHeaderLine;
 import org.molgenis.vcf.decisiontree.runner.info.GenotypeMetadataMapper;
+import org.molgenis.vcf.decisiontree.runner.info.MissingVepException;
 import org.molgenis.vcf.decisiontree.runner.info.NestedHeaderLine;
-import org.molgenis.vcf.decisiontree.runner.info.VepMetadataParser;
+import org.molgenis.vcf.decisiontree.runner.info.VepMetadataMapper;
 
 /**
  * {@link VCFFileReader} wrapper that works with nested metadata and data (e.g. CSQ INFO fields).
  */
 public class VcfReader implements AutoCloseable {
 
   private final VCFFileReader vcfFileReader;
-  private final VepMetadataParser vepMetadataParser;
+  private final VepMetadataMapper vepMetadataMapper;
   private final boolean strict;
   private final GenotypeMetadataMapper genotypeMetadataMapper;
   private boolean inited = false;
   private NestedHeaderLine vepNestedHeaderLine = null;
   private NestedHeaderLine gtNestedHeaderLine = null;
 
-  public VcfReader(VCFFileReader vcfFileReader, VepMetadataParser vepMetadataParser,
+  public VcfReader(VCFFileReader vcfFileReader, VepMetadataMapper vepMetadataMapper,
       GenotypeMetadataMapper genotypeMetadataMapper,
       boolean strict) {
     this.vcfFileReader = requireNonNull(vcfFileReader);
-    this.vepMetadataParser = requireNonNull(vepMetadataParser);
+    this.vepMetadataMapper = requireNonNull(vepMetadataMapper);
     this.genotypeMetadataMapper = requireNonNull(genotypeMetadataMapper);
     this.strict = strict;
   }
 
   private void initNestedMeta() {
     if (!inited) {
-      vepNestedHeaderLine = vepMetadataParser.map(vcfFileReader.getFileHeader());
+      String vepID = getVepId();
+      if(vepID == null) {
+        throw new MissingVepException();
+      }
+      vepNestedHeaderLine = vepMetadataMapper.map(vepID, vcfFileReader.getFileHeader());
       gtNestedHeaderLine = genotypeMetadataMapper.map();
       inited = true;
     }
   }
 
+  private String getVepId() {
+    for(VCFInfoHeaderLine vcfInfoHeaderLine : vcfFileReader.getFileHeader().getInfoHeaderLines()){
+      if(vepMetadataMapper.canMap(vcfInfoHeaderLine)){
+        return vcfInfoHeaderLine.getID();
+      }
+    }
+    return null;
+  }
+
   public Stream<VcfRecord> stream() {
     return StreamSupport.stream(vcfFileReader.spliterator(), false).map(VcfRecord::new);
   }

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/VcfRecord.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/VcfRecord.java
@@ -21,11 +21,10 @@
 import org.molgenis.vcf.decisiontree.filter.model.NestedField;
 import org.molgenis.vcf.decisiontree.filter.model.SampleContext;
 import org.molgenis.vcf.decisiontree.filter.model.SampleFieldType;
-import org.molgenis.vcf.decisiontree.filter.model.ValueCount;
-import org.molgenis.vcf.decisiontree.filter.model.ValueCount.Type;
-import org.molgenis.vcf.decisiontree.filter.model.ValueType;
 import org.molgenis.vcf.decisiontree.utils.VcfUtils;
 import org.molgenis.vcf.utils.UnexpectedEnumException;
+import org.molgenis.vcf.utils.metadata.ValueCount;
+import org.molgenis.vcf.utils.metadata.ValueType;
 import org.springframework.lang.Nullable;
 
 /**
@@ -240,7 +239,7 @@ private Object getExtendedAttributeValue(Field field, Genotype genotype) {
     }
 
     ValueCount valueCount = field.getValueCount();
-    Type valueCountType = valueCount.getType();
+    ValueCount.Type valueCountType = valueCount.getType();
     switch (valueCountType) {
       case A, R, VARIABLE:
         typedValue =
@@ -339,7 +338,7 @@ private Object getInfoValue(Field field, Allele allele) {
     Object value;
 
     ValueCount valueCount = field.getValueCount();
-    Type valueCountType = valueCount.getType();
+    ValueCount.Type valueCountType = valueCount.getType();
     switch (valueCountType) {
       case A:
         List<?> aInfoList = getInfoList(field);

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/model/Field.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/model/Field.java
@@ -1,5 +1,8 @@
 package org.molgenis.vcf.decisiontree.filter.model;
 
+import org.molgenis.vcf.utils.metadata.ValueCount;
+import org.molgenis.vcf.utils.metadata.ValueType;
+
 public interface Field {
 
   String getId();

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/model/FieldImpl.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/model/FieldImpl.java
@@ -1,21 +1,23 @@
 package org.molgenis.vcf.decisiontree.filter.model;
 
-import lombok.AllArgsConstructor;
-import lombok.Builder;
-import lombok.NonNull;
-import lombok.Value;
+import lombok.*;
 import lombok.experimental.NonFinal;
+import org.molgenis.vcf.utils.metadata.ValueCount;
+import org.molgenis.vcf.utils.metadata.ValueType;
 
 @Value
 @Builder
 @NonFinal
 @AllArgsConstructor
+@EqualsAndHashCode
 public class FieldImpl implements Field {
 
   @NonNull String id;
   @NonNull FieldType fieldType;
-  @NonNull ValueType valueType;
-  @NonNull ValueCount valueCount;
+  @NonNull
+  ValueType valueType;
+  @NonNull
+  ValueCount valueCount;
   /**
    * Returns count for FIXED value type, null otherwise
    */

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/model/GenotypeField.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/model/GenotypeField.java
@@ -6,6 +6,8 @@
 import lombok.NonNull;
 import lombok.Value;
 import lombok.experimental.NonFinal;
+import org.molgenis.vcf.utils.metadata.ValueCount;
+import org.molgenis.vcf.utils.metadata.ValueType;
 
 @Value
 @NonFinal

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/model/MissingField.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/model/MissingField.java
@@ -5,6 +5,8 @@
 import lombok.NonNull;
 import lombok.Value;
 import lombok.experimental.NonFinal;
+import org.molgenis.vcf.utils.metadata.ValueCount;
+import org.molgenis.vcf.utils.metadata.ValueType;
 
 @Value
 @Builder

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/model/NestedField.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/model/NestedField.java
@@ -3,12 +3,16 @@
 import static java.util.Objects.requireNonNull;
 
 import lombok.Builder;
+import lombok.EqualsAndHashCode;
 import lombok.NonNull;
 import lombok.Value;
 import lombok.experimental.NonFinal;
+import org.molgenis.vcf.utils.metadata.ValueCount;
+import org.molgenis.vcf.utils.metadata.ValueType;
 
 @Value
 @NonFinal
+@EqualsAndHashCode(callSuper = false)
 public class NestedField extends FieldImpl implements Comparable<NestedField> {
 
   final int index;

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/model/ValueCount.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/model/ValueCount.java
diff --git a/src/main/java/org/molgenis/vcf/decisiontree/filter/model/ValueType.java b/src/main/java/org/molgenis/vcf/decisiontree/filter/model/ValueType.java
diff --git a/src/main/java/org/molgenis/vcf/decisiontree/runner/QueryValidatorImpl.java b/src/main/java/org/molgenis/vcf/decisiontree/runner/QueryValidatorImpl.java
@@ -6,11 +6,11 @@
 import org.molgenis.vcf.decisiontree.filter.model.DecisionType;
 import org.molgenis.vcf.decisiontree.filter.model.Field;
 import org.molgenis.vcf.decisiontree.filter.model.MissingField;
-import org.molgenis.vcf.decisiontree.filter.model.ValueCount.Type;
-import org.molgenis.vcf.decisiontree.filter.model.ValueType;
 import org.molgenis.vcf.decisiontree.loader.model.ConfigBoolQuery;
 import org.molgenis.vcf.decisiontree.loader.model.ConfigOperator;
 import org.molgenis.vcf.utils.UnexpectedEnumException;
+import org.molgenis.vcf.utils.metadata.ValueCount;
+import org.molgenis.vcf.utils.metadata.ValueType;
 import org.springframework.stereotype.Component;
 
 import static org.molgenis.vcf.decisiontree.loader.model.ConfigOperator.*;
@@ -79,17 +79,17 @@ private void validateContains(Field field, ConfigBoolQuery configBoolQuery) {
     validateFileValueAllowed(configBoolQuery, field);
     if (field.getValueType() == ValueType.FLAG || field.getValueType() == ValueType.FLOAT) {
       throw new UnsupportedValueTypeException(field, DecisionType.BOOL);
-    } else if (field.getValueCount().getType() == Type.FIXED
+    } else if (field.getValueCount().getType() == ValueCount.Type.FIXED
         && field.getValueCount().getCount() == 1) {
       throw new UnsupportedValueCountException(field, DecisionType.BOOL, CONTAINS);
     }
   }
 
   private void validateSingleOrPerAlleleCount(Field field, DecisionType decisionType, ConfigOperator configOperator) {
-    if (field.getValueCount().getType() == Type.G ||
-        field.getValueCount().getType() == Type.VARIABLE) {
+    if (field.getValueCount().getType() == ValueCount.Type.G ||
+        field.getValueCount().getType() == ValueCount.Type.VARIABLE) {
       throw new UnsupportedValueCountTypeException(field, decisionType, configOperator);
-    } else if (field.getValueCount().getType() == Type.FIXED
+    } else if (field.getValueCount().getType() == ValueCount.Type.FIXED
         && field.getValueCount().getCount() != 1) {
       throw new UnsupportedValueCountException(field, decisionType, configOperator);
     }

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/runner/ValueValidator.java b/src/main/java/org/molgenis/vcf/decisiontree/runner/ValueValidator.java
@@ -19,7 +19,6 @@
 import org.molgenis.vcf.decisiontree.filter.VcfMetadata;
 import org.molgenis.vcf.decisiontree.filter.model.Field;
 import org.molgenis.vcf.decisiontree.filter.model.MissingField;
-import org.molgenis.vcf.decisiontree.filter.model.ValueType;
 import org.molgenis.vcf.decisiontree.loader.ConfigDecisionTreeValidationException;
 import org.molgenis.vcf.decisiontree.loader.model.ConfigBoolMultiNode;
 import org.molgenis.vcf.decisiontree.loader.model.ConfigBoolNode;
@@ -28,6 +27,7 @@
 import org.molgenis.vcf.decisiontree.loader.model.ConfigNode;
 import org.molgenis.vcf.decisiontree.loader.model.ConfigOperator;
 import org.molgenis.vcf.utils.UnexpectedEnumException;
+import org.molgenis.vcf.utils.metadata.ValueType;
 import org.molgenis.vcf.utils.sample.model.AffectedStatus;
 import org.molgenis.vcf.utils.sample.model.Sex;
 

diff --git a/src/main/java/org/molgenis/vcf/decisiontree/runner/VcfReaderFactoryImpl.java b/src/main/java/org/molgenis/vcf/decisiontree/runner/VcfReaderFactoryImpl.java
@@ -7,29 +7,29 @@
 import org.molgenis.vcf.decisiontree.Settings;
 import org.molgenis.vcf.decisiontree.filter.VcfReader;
 import org.molgenis.vcf.decisiontree.runner.info.GenotypeMetadataMapper;
-import org.molgenis.vcf.decisiontree.runner.info.VepMetadataParser;
-import org.molgenis.vcf.decisiontree.runner.info.VepMetadataParserFactory;
+import org.molgenis.vcf.decisiontree.runner.info.VepMetadataMapper;
+import org.molgenis.vcf.decisiontree.runner.info.VepMetadataMapperFactory;
 import org.springframework.stereotype.Component;
 
 @Component
 class VcfReaderFactoryImpl implements VcfReaderFactory {
 
-  private final VepMetadataParserFactory vepMetadataParserFactory;
   private final GenotypeMetadataMapper genotypeMetadataMapper;
+  private final VepMetadataMapperFactory vepMetadataMapperFactory;
 
-  VcfReaderFactoryImpl(VepMetadataParserFactory vepMetadataParserFactory,
+  VcfReaderFactoryImpl(VepMetadataMapperFactory vepMetadataMapperFactory,
                        GenotypeMetadataMapper genotypeMetadataMapper) {
-    this.vepMetadataParserFactory = requireNonNull(vepMetadataParserFactory);
+    this.vepMetadataMapperFactory = requireNonNull(vepMetadataMapperFactory);
     this.genotypeMetadataMapper = requireNonNull(genotypeMetadataMapper);
   }
 
   @Override
   public VcfReader create(Settings settings) {
-    VepMetadataParser vepMetadataParser = vepMetadataParserFactory.create(settings);
+    VepMetadataMapper vepMetadataMapper = vepMetadataMapperFactory.create(settings);
 
     Path inputVcfPath = settings.getInputVcfPath();
     boolean strict = settings.isStrict();
-    return new VcfReader(new VCFFileReader(inputVcfPath.toFile(), false), vepMetadataParser,
+    return new VcfReader(new VCFFileReader(inputVcfPath.toFile(), false), vepMetadataMapper,
         genotypeMetadataMapper,
         strict);
   }