diff --git a/protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/variantutils/VariantsToTableIntegrationTest.java b/protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/variantutils/VariantsToTableIntegrationTest.java
index ed937c881..8d5fadbb5 100644
--- a/protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/variantutils/VariantsToTableIntegrationTest.java
+++ b/protected/gatk-tools-protected/src/test/java/org/broadinstitute/gatk/tools/walkers/variantutils/VariantsToTableIntegrationTest.java
@@ -63,7 +63,7 @@ public class VariantsToTableIntegrationTest extends WalkerTest {
" --variant:vcf " + privateTestDir + "soap_gatk_annotated.vcf" +
" -T VariantsToTable" +
" -F CHROM -F POS -F ID -F REF -F ALT -F QUAL -F FILTER -F TRANSITION -F DP -F SB -F set -F RankSumP -F refseq.functionalClass*" +
- " -L chr1 -o %s" + moreArgs;
+ " -L chr1 -o %s " + moreArgs;
}
private String variantsToTableMultiAllelicCmd(String moreArgs) {
@@ -76,15 +76,15 @@ public class VariantsToTableIntegrationTest extends WalkerTest {
@Test(enabled = true)
public void testComplexVariantsToTable() {
- WalkerTestSpec spec = new WalkerTestSpec(variantsToTableCmd(" -AMD"),
+ WalkerTestSpec spec = new WalkerTestSpec(variantsToTableCmd(""),
Arrays.asList("e8f771995127b727fb433da91dd4ee98"));
executeTest("testComplexVariantsToTable", spec);
}
@Test(enabled = true)
public void testComplexVariantsToTableFail() {
- WalkerTestSpec spec = new WalkerTestSpec(variantsToTableCmd(""), 1, UserException.class);
- executeTest("testComplexVariantsToTable-FAIL", spec);
+ WalkerTestSpec spec = new WalkerTestSpec(variantsToTableCmd("--errorIfMissingData"), 1, UserException.class);
+ executeTest("testComplexVariantsToTableFail", spec);
}
@Test(enabled = true)
@@ -114,6 +114,32 @@ public class VariantsToTableIntegrationTest extends WalkerTest {
executeTest("testGenotypeFields", spec);
}
+ @Test(enabled = true)
+ public void testUnfilteredGenotypeField() {
+ WalkerTest.WalkerTestSpec spec = new WalkerTest.WalkerTestSpec(
+ "-R " + b36KGReference +
+ " --variant " + privateTestDir + "vcfexample2.vcf" +
+ " -T VariantsToTable" +
+ " -GF RD -GF FT" +
+ " -o %s",
+ 1,
+ Arrays.asList("fec1e70f3c5762225535559feb5ffdee"));
+ executeTest("testUnfilteredGenotypeField", spec);
+ }
+
+ @Test(enabled = true)
+ public void testUnfilteredGenotypeFieldFail() {
+ WalkerTest.WalkerTestSpec spec = new WalkerTest.WalkerTestSpec(
+ "-R " + b36KGReference +
+ " --variant " + privateTestDir + "vcfexample2.vcf" +
+ " -T VariantsToTable" +
+ " -GF RD -GF FT --errorIfMissingData" +
+ " -o %s",
+ 1,
+ UserException.class);
+ executeTest("testUnfilteredGenotypeFieldFail", spec);
+ }
+
@Test
public void testMultiallelicGenotypeFields() {
WalkerTest.WalkerTestSpec spec = new WalkerTest.WalkerTestSpec(
diff --git a/public/gatk-tools-public/src/main/java/org/broadinstitute/gatk/tools/walkers/variantutils/VariantsToTable.java b/public/gatk-tools-public/src/main/java/org/broadinstitute/gatk/tools/walkers/variantutils/VariantsToTable.java
index 5a4b7e2b5..5ecbae794 100644
--- a/public/gatk-tools-public/src/main/java/org/broadinstitute/gatk/tools/walkers/variantutils/VariantsToTable.java
+++ b/public/gatk-tools-public/src/main/java/org/broadinstitute/gatk/tools/walkers/variantutils/VariantsToTable.java
@@ -98,7 +98,7 @@ import java.util.*;
*
* Caveats
*
- * - Some annotations cannot be applied to all variant sites, so VCFs typically contain records where some annotation values are missing. By default this tool throws an error if you request export of an annotation for which not all records have values. You can override this behavior by setting --allowMissingData in the command line. As a result, the tool will emit the special value NA for the missing annotations in those records.
+ * - Some annotations cannot be applied to all variant sites, so VCFs typically contain records where some annotation values are missing. By default this tool the tool will emit the special value NA for the missing annotations if you request export of an annotation for which not all records have values. You can override this behavior by setting --errorIfMissingData in the command line. As a result, the tool will throw an error if a record is missing a value.
* - When you request export of sample-level annotations (FORMAT field annotations such as GT), the annotations will be identified per-sample. If multiple samples are present in the VCF, the columns will be ordered alphabetically by sample name (SM tag).
*
*
@@ -173,18 +173,28 @@ public class VariantsToTable extends RodWalker {
* fields (e.g., AC not being calculated for filtered records, if included). When provided, this argument
* will cause VariantsToTable to write out NA values for missing fields instead of throwing an error.
*/
- @Advanced
- @Argument(fullName="allowMissingData", shortName="AMD", doc="If provided, we will not require every record to contain every field", required=false)
+ @Hidden
+ @Deprecated
+ @Argument(fullName="allowMissingData", shortName="AMD", doc="This argument is no longer used in GATK versions 3.8 and newer. " +
+ "Please see the online documentation for --errorIfMissingData.", required=false)
public boolean ALLOW_MISSING_DATA = false;
+
+ /**
+ * By default, this tool will write out NA values indicating missing data when it encounters a field without a value in a record.
+ * If this flag is added to the command, the tool will instead exit with an error if missing data is encountered..
+ */
+ @Advanced
+ @Argument(fullName="errorIfMissingData", shortName="EMD", doc="If provided, we will require every record to contain every field", required=false)
+ public boolean ERROR_IF_MISSING_DATA = false;
private final static String MISSING_DATA = "NA";
- private final List samples = new ArrayList();
+ private final List samples = new ArrayList<>();
public void initialize() {
if ( !genotypeFieldsToTake.isEmpty() ) {
Map vcfRods = GATKVCFUtils.getVCFHeadersFromRods(getToolkit(), variants);
- TreeSet vcfSamples = new TreeSet(SampleUtils.getSampleList(vcfRods, GATKVariantContextUtils.GenotypeMergeType.REQUIRE_UNIQUE));
+ TreeSet vcfSamples = new TreeSet<>(SampleUtils.getSampleList(vcfRods, GATKVariantContextUtils.GenotypeMergeType.REQUIRE_UNIQUE));
samples.addAll(vcfSamples);
// optimization: if there are no samples, we don't have to worry about any genotype fields
@@ -210,7 +220,7 @@ public class VariantsToTable extends RodWalker {
for ( VariantContext vc : tracker.getValues(variants, context.getLocation())) {
if ( showFiltered || vc.isNotFiltered() ) {
nRecords++;
- for ( final List record : extractFields(vc, fieldsToTake, genotypeFieldsToTake, samples, ALLOW_MISSING_DATA, splitMultiAllelic) ) {
+ for ( final List record : extractFields(vc, fieldsToTake, genotypeFieldsToTake, samples, !ERROR_IF_MISSING_DATA, splitMultiAllelic) ) {
if ( moltenizeOutput )
emitMoltenizedOutput(record);
else
@@ -281,7 +291,7 @@ public class VariantsToTable extends RodWalker {
final boolean splitMultiAllelic) {
final int numRecordsToProduce = splitMultiAllelic ? vc.getAlternateAlleles().size() : 1;
- final List> records = new ArrayList>(numRecordsToProduce);
+ final List> records = new ArrayList<>(numRecordsToProduce);
int numFields = fields.size();
final boolean addGenotypeFields = genotypeFields != null && !genotypeFields.isEmpty();
@@ -289,7 +299,7 @@ public class VariantsToTable extends RodWalker {
numFields += genotypeFields.size() * samples.size();
for ( int i = 0; i < numRecordsToProduce; i++ )
- records.add(new ArrayList(numFields));
+ records.add(new ArrayList<>(numFields));
for ( String field : fields ) {
@@ -302,7 +312,7 @@ public class VariantsToTable extends RodWalker {
} else if ( vc.hasAttribute(field) ) {
addFieldValue(vc.getAttribute(field, null), records);
} else if ( isWildCard(field) ) {
- Set wildVals = new HashSet();
+ Set wildVals = new HashSet<>();
for ( Map.Entry elt : vc.getAttributes().entrySet()) {
if ( elt.getKey().startsWith(field.substring(0, field.length() - 1)) ) {
wildVals.add(elt.getValue().toString());
@@ -310,17 +320,15 @@ public class VariantsToTable extends RodWalker {
}
String val = MISSING_DATA;
- if ( wildVals.size() > 0 ) {
- List toVal = new ArrayList(wildVals);
+ if ( !wildVals.isEmpty() ) {
+ List toVal = new ArrayList<>(wildVals);
Collections.sort(toVal);
val = Utils.join(",", toVal);
}
addFieldValue(val, records);
- } else if ( ! allowMissingData ) {
- throw new UserException(String.format("Missing field %s in vc %s at %s", field, vc.getSource(), vc));
} else {
- addFieldValue(MISSING_DATA, records);
+ handleMissingData(allowMissingData, field, records, vc);
}
}
@@ -330,11 +338,22 @@ public class VariantsToTable extends RodWalker {
if ( vc.hasGenotype(sample) && vc.getGenotype(sample).hasAnyAttribute(gf) ) {
if ( gf.equals(VCFConstants.GENOTYPE_KEY) )
addFieldValue(vc.getGenotype(sample).getGenotypeString(true), records);
- else
- addFieldValue(vc.getGenotype(sample).getAnyAttribute(gf), records);
+ else {
+ /**
+ * TODO - If gf == "FT" and the GT record is not filtered, Genotype.getAnyAttribute == null. Genotype.hasAnyAttribute should be changed so it
+ * returns false for this condition. Presently, it always returns true. Once this is fixed, then only the "addFieldValue" statement will
+ * remain in the following logic block.
+ */
+ if (vc.getGenotype(sample).getAnyAttribute(gf) != null) {
+ addFieldValue(vc.getGenotype(sample).getAnyAttribute(gf), records);
+ } else {
+ handleMissingData(allowMissingData, gf, records, vc);
+ }
+ }
+ }
+ else {
+ handleMissingData(allowMissingData, gf, records, vc);
}
- else
- addFieldValue(MISSING_DATA, records);
}
}
}
@@ -342,9 +361,17 @@ public class VariantsToTable extends RodWalker {
return records;
}
+ private static void handleMissingData(final boolean allowMissingData, final String field, final List> records, final VariantContext vc) {
+ if (allowMissingData) {
+ addFieldValue(MISSING_DATA, records);
+ } else {
+ throw new UserException(String.format("Missing field %s in vc %s at %s", field, vc.getSource(), vc));
+ }
+ }
+
private static void addFieldValue(final Object val, final List> result) {
final int numResultRecords = result.size();
-
+
// if we're trying to create a single output record, add it
if ( numResultRecords == 1 ) {
result.get(0).add(prettyPrintObject(val));
@@ -364,6 +391,10 @@ public class VariantsToTable extends RodWalker {
}
private static String prettyPrintObject(final Object val) {
+ // should never occur
+ if (val == null)
+ return "";
+
if ( val instanceof List )
return prettyPrintObject(((List)val).toArray());
@@ -400,7 +431,7 @@ public class VariantsToTable extends RodWalker {
// ----------------------------------------------------------------------------------------------------
public static abstract class Getter { public abstract String get(VariantContext vc); }
- public static final Map getters = new HashMap();
+ public static final Map getters = new HashMap<>();
static {
// #CHROM POS ID REF ALT QUAL FILTER INFO FORMAT