Merge pull request #643 from CAWarmerdam/master

Bug fixes and minor improvements to Genotype-IO and GH

Genotype-Harmonizer/pom.xml

@@ -7,7 +7,7 @@
 	</parent>
 	<modelVersion>4.0.0</modelVersion>
 	<artifactId>Genotype-Harmonizer</artifactId>
-	<version>1.4.25-SNAPSHOT</version>
+	<version>1.4.26-SNAPSHOT</version>
 	<name>Genotype Harmonizer</name>
 	<packaging>jar</packaging>
 	<dependencies>

Genotype-Harmonizer/src/main/java/nl/umcg/deelenp/genotypeharmonizer/GenotypeHarmonizerParamaters.java

@@ -441,20 +441,20 @@ public GenotypeHarmonizerParamaters(String... args) throws ParseException {
                         + "\" is not a supported genotype field.");
             }
 
-            boolean raiseExceptionIfUnavailable = true;
+            boolean forcePreferredGenotypeFormat = true;
             if (genotypeFormatArguments.length > 2) {
                 if (genotypeFormatArguments[2].equals("suppress")) {
                     System.out.println("WARNING: requested to supress exceptions if preferred genotype format is unavailable. For those variants the default will be chosen.");
                     LOGGER.warn("WARNING: requested to supress exceptions if preferred genotype format is unavailable. For those variants the default will be chosen.");
-                    raiseExceptionIfUnavailable = false;
+                    forcePreferredGenotypeFormat = false;
                 }
             }
 
             nonFinalVcfGenotypeFormatSupplier = new VcfGenotypeFormatSupplier(
                     VcfGenotypeFormat.valueOf(genotypeFormatArguments[0]),
                     genotypeFormatArguments.length > 1 ?
                             genotypeFormatArguments[1] : genotypeFormatArguments[0],
-                    raiseExceptionIfUnavailable);
+                    forcePreferredGenotypeFormat);
         }
 
         vcfGenotypeFormatSupplier = nonFinalVcfGenotypeFormatSupplier;

Genotype-IO/src/main/java/org/molgenis/genotype/bgen/BgenGenotypeWriter.java

@@ -301,8 +301,7 @@ private void addMetaData(File bgenFile, BgenixWriter bgenixWriter) throws IOExce
 		byte[] firstBytes = new byte[1000];
 		randomAccessBgenFile.read(firstBytes, 0, 1000);
 
-		//Add current time in int.
-		System.out.println((System.currentTimeMillis() / 1000L));
+		// Add current time in int.
 		// Create and write new metadata.
 		BgenixMetadata m = new BgenixMetadata(
 				bgenFile.getName(),

Genotype-IO/src/main/java/org/molgenis/genotype/vcf/VcfGenotypeData.java

@@ -65,6 +65,7 @@ public class VcfGenotypeData extends AbstractRandomAccessGenotypeData implements
     private final LinkedHashSet<VcfGenotypeFormat> genotypeProbabilitiesFieldPrecedence;
     private final LinkedHashSet<VcfGenotypeFormat> genotypeCallFieldPrecedence;
     private final LinkedHashSet<VcfGenotypeFormat> genotypeDosageFieldPrecedence;
+    private final LinkedHashSet<VcfGenotypeFormat> haplotypeProbabilitiesFieldPresedence;
 
 
     /**
@@ -138,6 +139,8 @@ public VcfGenotypeData(File bzipVcfFile, File tabixIndexFile, int cacheSize, dou
                 new LinkedHashSet<>(Arrays.asList(VcfGenotypeFormat.GT, VcfGenotypeFormat.GP, VcfGenotypeFormat.DS));
         genotypeDosageFieldPrecedence =
                 new LinkedHashSet<>(Arrays.asList(VcfGenotypeFormat.DS, VcfGenotypeFormat.GP, VcfGenotypeFormat.GT));
+        haplotypeProbabilitiesFieldPresedence =
+                new LinkedHashSet<>(Arrays.asList(VcfGenotypeFormat.HP, VcfGenotypeFormat.ADS));
 
         genotypeFormatSupplier = new VcfGenotypeFormatSupplier();
     }
@@ -321,15 +324,12 @@ public boolean arePhasedProbabilitiesPresent(GeneticVariant variant) {
         LinkedHashSet<VcfGenotypeFormat> haplotypeProbabilitiesFields = getVcfHaplotypeFormats(variant);
 
         // If the requested format is set and present for this variant base decision on this format
-        VcfGenotypeFormat genotypeFormat = genotypeFormatSupplier.getVcfGenotypeFormat(
+        return genotypeFormatSupplier.vcfGenotypeFormatReadable(
                 vcfRecord, haplotypeProbabilitiesFields);
-
-        return (genotypeFormat != null);
     }
 
     private LinkedHashSet<VcfGenotypeFormat> getVcfHaplotypeFormats(GeneticVariant variant) {
-        LinkedHashSet<VcfGenotypeFormat> haplotypeProbabilitiesFields =
-                new LinkedHashSet<>(Arrays.asList(VcfGenotypeFormat.HP, VcfGenotypeFormat.ADS));
+        LinkedHashSet<VcfGenotypeFormat> haplotypeProbabilitiesFields = haplotypeProbabilitiesFieldPresedence;
 
         if (variant.hasPhasedGenotypes()) {
             haplotypeProbabilitiesFields.add(VcfGenotypeFormat.GT);

Genotype-IO/src/main/java/org/molgenis/genotype/vcf/VcfGenotypeField/VcfGenotypeFormatSupplier.java

@@ -1,6 +1,5 @@
 package org.molgenis.genotype.vcf.VcfGenotypeField;
 
-import org.apache.commons.lang3.StringUtils;
 import org.molgenis.genotype.GenotypeDataException;
 import org.molgenis.vcf.VcfRecord;
 
@@ -15,7 +14,7 @@
 public class VcfGenotypeFormatSupplier {
     private VcfGenotypeFormat preferredGenotypeFormat;
     private String preferredGenotypeFormatIdentifier;
-    private boolean raiseExceptionIfUnavailable;
+    private boolean forcePreferredGenotypeFormat;
 
     public VcfGenotypeFormatSupplier(VcfGenotypeFormat preferredGenotypeFormat) {
         this(preferredGenotypeFormat, preferredGenotypeFormat.toString(), false);
@@ -29,11 +28,11 @@ public VcfGenotypeFormatSupplier(VcfGenotypeFormat preferredGenotypeFormat, bool
         this(preferredGenotypeFormat, preferredGenotypeFormat.toString(), false);
     }
 
-    public VcfGenotypeFormatSupplier(VcfGenotypeFormat preferredGenotypeFormat, String formatIdentifier, boolean raiseExceptionIfUnavailable) {
+    public VcfGenotypeFormatSupplier(VcfGenotypeFormat preferredGenotypeFormat, String formatIdentifier, boolean forcePreferredGenotypeFormat) {
 
         this.preferredGenotypeFormat = preferredGenotypeFormat;
         this.preferredGenotypeFormatIdentifier = formatIdentifier;
-        this.raiseExceptionIfUnavailable = raiseExceptionIfUnavailable;
+        this.forcePreferredGenotypeFormat = forcePreferredGenotypeFormat;
     }
 
     public VcfGenotypeFormatSupplier() {
@@ -55,31 +54,96 @@ public VcfGenotypeFormat getVcfGenotypeFormat(
 
         List<String> formatIdentifiers = Arrays.asList(vcfRecord.getFormat());
 
-        if (preferredGenotypeFormat != null
-                && genotypeDosageFieldPrecedence.contains(preferredGenotypeFormat)
-                && formatIdentifiers.contains(this.getGenotypeFormatIdentifier(preferredGenotypeFormat))) {
-            return preferredGenotypeFormat;
-        }
-
-        if (this.raiseExceptionIfUnavailable) {
-            throw new GenotypeDataException(String.format(
-                    "Preferred genotype format field (%s) is unavailable for vcf record: %n%s (%s:%s). " +
-                            "Available format fields: %s",
-                    preferredGenotypeFormatIdentifier,
-                    String.join(", ", vcfRecord.getIdentifiers()),
-                    vcfRecord.getChromosome(), vcfRecord.getPosition(),
-                    String.join(", ", vcfRecord.getFormat())));
+        // Check if the preferred genotype format is set
+        if (preferredGenotypeFormat != null) {
+            // If it is set, check if it is available, and, if it is not, if we should write exceptions or not.
+            if (genotypeDosageFieldPrecedence.contains(preferredGenotypeFormat)
+                    && isGenotypeFormatPresent(formatIdentifiers, preferredGenotypeFormat)) {
+                return preferredGenotypeFormat;
+            } else if (this.forcePreferredGenotypeFormat) {
+                if (!isGenotypeFormatPresent(formatIdentifiers, preferredGenotypeFormat)) {
+                    throw new GenotypeDataException(String.format(
+                            "Preferred genotype format field (%s) is unavailable for vcf record: %n%s (%s:%s). " +
+                                    "Available format fields: %s",
+                            preferredGenotypeFormatIdentifier,
+                            String.join(", ", vcfRecord.getIdentifiers()),
+                            vcfRecord.getChromosome(), vcfRecord.getPosition(),
+                            String.join(", ", vcfRecord.getFormat())));
+                } else if (!genotypeDosageFieldPrecedence.contains(preferredGenotypeFormat)) {
+                    throw new GenotypeDataException(String.format(
+                            "Preferred genotype format field (%s) cannot be used. " +
+                                    "Requested to load vcf record %n%s (%s:%s). " +
+                                    "Possible format fields: %s",
+                            preferredGenotypeFormatIdentifier,
+                            String.join(", ", vcfRecord.getIdentifiers()),
+                            vcfRecord.getChromosome(), vcfRecord.getPosition(),
+                            String.join(", ", Arrays.toString(genotypeDosageFieldPrecedence.toArray()))));
+                }
+            }
         }
 
         for (VcfGenotypeFormat genotypeFormat: genotypeDosageFieldPrecedence) {
-            if (formatIdentifiers.contains(this.getGenotypeFormatIdentifier(genotypeFormat))) {
+            if (isGenotypeFormatPresent(formatIdentifiers, genotypeFormat)) {
                 return genotypeFormat;
             }
         }
 
         return null;
     }
 
+    /**
+     * @param vcfRecord record, row, within a VCF file. corresponding to a particular variant.
+     * @param genotypeDosageFieldPrecedence LinkedHashSet that lists all formats that can be read,
+     *                                      in order of precedence (high precedence to low precedence).
+     * @return If there is a preferred genotype format supplied, this method only returns true if the
+     * preferred genotype format is available from the vcf record and the list of
+     * possible formats that can be read according to the genotype field precedence hash set.
+     * If a preferred genotype format is not supplied, this method will return true if one of
+     * the genotype field formats from the precedence list can be read from the vcf record.
+     * If nothing matches these conditions, false is returned.
+     */
+    public boolean vcfGenotypeFormatReadable(
+            VcfRecord vcfRecord,
+            LinkedHashSet<VcfGenotypeFormat> genotypeDosageFieldPrecedence) {
+
+        List<String> formatIdentifiers = Arrays.asList(vcfRecord.getFormat());
+
+        // Test if the preferred genotype format is present
+        // Test if we should suppress exception if this is not the case
+        // Test if the
+
+        if (preferredGenotypeFormat != null) {
+            if (isGenotypeFormatPresent(formatIdentifiers, preferredGenotypeFormat)) {
+                return genotypeDosageFieldPrecedence.contains(preferredGenotypeFormat);
+            } else if (this.forcePreferredGenotypeFormat) {
+                throw new GenotypeDataException(String.format(
+                        "Preferred genotype format field (%s) is unavailable for vcf record: %n%s (%s:%s). " +
+                                "Available format fields: %s",
+                        preferredGenotypeFormatIdentifier,
+                        String.join(", ", vcfRecord.getIdentifiers()),
+                        vcfRecord.getChromosome(), vcfRecord.getPosition(),
+                        String.join(", ", vcfRecord.getFormat())));
+            }
+        }
+
+        for (VcfGenotypeFormat genotypeFormat: genotypeDosageFieldPrecedence) {
+            if (isGenotypeFormatPresent(formatIdentifiers, genotypeFormat)) {
+                return true;
+            }
+        }
+
+        return false;
+    }
+
+    public boolean isPreferredGenotypeFormatPresent(VcfRecord vcfRecord) {
+        List<String> formatIdentifiers = Arrays.asList(vcfRecord.getFormat());
+        return isGenotypeFormatPresent(formatIdentifiers, preferredGenotypeFormat);
+    }
+
+    private boolean isGenotypeFormatPresent(List<String> formatIdentifiers, VcfGenotypeFormat genotypeFormat) {
+        return formatIdentifiers.contains(this.getGenotypeFormatIdentifier(genotypeFormat));
+    }
+
     public VcfGenotypeFormat getPreferredGenotypeFormat() {
         return preferredGenotypeFormat;
     }