Skip to content

Commit

Permalink
Changing the name of the GencodeFuncotationFactory to be read from file.
Browse files Browse the repository at this point in the history 
Fixes #3956

Now gencode data sources have names preserved from config files.

Updated MafOutputRenderer to put a space and delimiter between the date and first funcotation factory information.
Updated some test cases to be correct with the new Gencode name preservation and MAF renderer update.
  • Loading branch information
@jonn-smith
jonn-smith committed May 29, 2018
1 parent 70adf91 commit 87b6ce8
Show file tree
Hide file tree
Showing 11 changed files with 121 additions and 68 deletions.
2 changes: 2 additions & 0 deletions ...main/java/org/broadinstitute/hellbender/tools/funcotator/dataSources/DataSourceUtils.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -392,10 +392,12 @@ public static GencodeFuncotationFactory createGencodeDataSource(final Path dataS
// Get some metadata:
final String fastaPath = dataSourceProperties.getProperty(CONFIG_FILE_FIELD_NAME_GENCODE_FASTA_PATH);
final String version = dataSourceProperties.getProperty(CONFIG_FILE_FIELD_NAME_VERSION);
final String name = dataSourceProperties.getProperty(CONFIG_FILE_FIELD_NAME_NAME);

// Create our gencode factory:
return new GencodeFuncotationFactory(dataSourceFile.resolveSibling(fastaPath),
version,
name,
transcriptSelectionMode,
userTranscriptIdSet,
annotationOverridesMap
Expand Down
65 changes: 38 additions & 27 deletions ...rg/broadinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotation.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -14,7 +14,7 @@
import java.util.stream.Collectors;

/**
* A class to represent a Functional Annotation.
* A class to represent a Functional Annotation from the Gencode data source.
* Created by jonn on 8/22/17.
*/
public class GencodeFuncotation implements Funcotation {
Expand Down Expand Up @@ -53,6 +53,8 @@ public class GencodeFuncotation implements Funcotation {

private List<String> otherTranscripts; // TRIVIAL

private String dataSourceName;

//------------------------------------------------------------
// Non-serialized fields:

Expand Down Expand Up @@ -121,6 +123,7 @@ public GencodeFuncotation(final GencodeFuncotation that) {
this.gcContent = that.gcContent;
this.referenceContext = that.referenceContext;
this.otherTranscripts = that.otherTranscripts;
this.dataSourceName = that.dataSourceName;
this.locusLevel = that.locusLevel;
this.apprisRank = that.apprisRank;
this.transcriptLength = that.transcriptLength;
Expand Down Expand Up @@ -201,7 +204,7 @@ public String serializeToVcfString() {
public void setFieldSerializationOverrideValue( final String fieldName, final String overrideValue ) {

// Cut off the "Gencode" and version number at the start of the string:
final String shortFieldName = fieldName.replaceAll("^" + GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_", "");
final String shortFieldName = fieldName.replaceAll("^" + getDataSourceName()+ "_" + version + "_", "");

switch (shortFieldName) {
case "hugoSymbol": hugoSymbolSerializedOverride = overrideValue; break;
Expand Down Expand Up @@ -232,35 +235,35 @@ public void setFieldSerializationOverrideValue( final String fieldName, final St

@Override
public String getDataSourceName() {
return GencodeFuncotationFactory.DATA_SOURCE_NAME;
return dataSourceName;
}

@Override
public LinkedHashSet<String> getFieldNames() {
return new LinkedHashSet<>(
Arrays.asList(
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_hugoSymbol",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_ncbiBuild",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_chromosome",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_start",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_end",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_variantClassification",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_secondaryVariantClassification",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_variantType",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_refAllele",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_tumorSeqAllele1",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_tumorSeqAllele2",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_genomeChange",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_annotationTranscript",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_transcriptStrand",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_transcriptExon",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_transcriptPos",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_cDnaChange",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_codonChange",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_proteinChange",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_gcContent",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_referenceContext",
GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_otherTranscripts"
getDataSourceName() + "_" + version + "_hugoSymbol",
getDataSourceName() + "_" + version + "_ncbiBuild",
getDataSourceName() + "_" + version + "_chromosome",
getDataSourceName() + "_" + version + "_start",
getDataSourceName() + "_" + version + "_end",
getDataSourceName() + "_" + version + "_variantClassification",
getDataSourceName() + "_" + version + "_secondaryVariantClassification",
getDataSourceName() + "_" + version + "_variantType",
getDataSourceName() + "_" + version + "_refAllele",
getDataSourceName() + "_" + version + "_tumorSeqAllele1",
getDataSourceName() + "_" + version + "_tumorSeqAllele2",
getDataSourceName() + "_" + version + "_genomeChange",
getDataSourceName() + "_" + version + "_annotationTranscript",
getDataSourceName() + "_" + version + "_transcriptStrand",
getDataSourceName() + "_" + version + "_transcriptExon",
getDataSourceName() + "_" + version + "_transcriptPos",
getDataSourceName() + "_" + version + "_cDnaChange",
getDataSourceName() + "_" + version + "_codonChange",
getDataSourceName() + "_" + version + "_proteinChange",
getDataSourceName() + "_" + version + "_gcContent",
getDataSourceName() + "_" + version + "_referenceContext",
getDataSourceName() + "_" + version + "_otherTranscripts"
)
);
}
Expand All @@ -270,11 +273,11 @@ public String getField(final String fieldName) {

// Allow a user to specify the name of the field, or the fully-qualified name of the field
// with GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_" at the start.
final String altFieldName = GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_" + fieldName;
final String altFieldName = getDataSourceName() + "_" + version + "_" + fieldName;
final LinkedHashSet<String> fieldNames = getFieldNames();

if ( fieldNames.contains(fieldName) || fieldNames.contains(altFieldName) ) {
switch(fieldName.replace(GencodeFuncotationFactory.DATA_SOURCE_NAME + "_" + version + "_", "")) {
switch(fieldName.replace(getDataSourceName() + "_" + version + "_", "")) {
case "hugoSymbol":
return (hugoSymbolSerializedOverride != null ? hugoSymbolSerializedOverride : (hugoSymbol != null ? hugoSymbol : ""));
case "ncbiBuild":
Expand Down Expand Up @@ -363,6 +366,8 @@ public boolean equals(final Object o) {
return false;
if ( otherTranscripts != null ? !otherTranscripts.equals(that.otherTranscripts) : that.otherTranscripts != null )
return false;
if ( dataSourceName != null ? !dataSourceName.equals(that.dataSourceName) : that.dataSourceName != null )
return false;
if ( locusLevel != null ? !locusLevel.equals(that.locusLevel) : that.locusLevel != null ) return false;
if ( apprisRank != that.apprisRank ) return false;
if ( transcriptLength != null ? !transcriptLength.equals(that.transcriptLength) : that.transcriptLength != null )
Expand Down Expand Up @@ -437,6 +442,7 @@ public int hashCode() {
result = 31 * result + (gcContent != null ? gcContent.hashCode() : 0);
result = 31 * result + (referenceContext != null ? referenceContext.hashCode() : 0);
result = 31 * result + (otherTranscripts != null ? otherTranscripts.hashCode() : 0);
result = 31 * result + (dataSourceName != null ? dataSourceName.hashCode() : 0);
result = 31 * result + (locusLevel != null ? locusLevel.hashCode() : 0);
result = 31 * result + (apprisRank != null ? apprisRank.hashCode() : 0);
result = 31 * result + (transcriptLength != null ? transcriptLength.hashCode() : 0);
Expand Down Expand Up @@ -491,6 +497,7 @@ public String toString() {
", gcContent=" + gcContent +
", referenceContext='" + referenceContext + '\'' +
", otherTranscripts=" + otherTranscripts +
", dataSourceName=" + dataSourceName +
", locusLevel=" + locusLevel +
", apprisRank=" + apprisRank +
", transcriptLength=" + transcriptLength +
Expand Down Expand Up @@ -726,6 +733,10 @@ public void setGeneTranscriptType(final GencodeGtfFeature.GeneTranscriptType gen
this.geneTranscriptType = geneTranscriptType;
}

public void setDataSourceName(final String dataSourceName) {
this.dataSourceName = dataSourceName;
}

//==================================================================================================================

/**
Expand Down
10 changes: 10 additions & 0 deletions ...dinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationBuilder.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -298,4 +298,14 @@ public GencodeFuncotationBuilder setGeneTranscriptType(final GencodeGtfFeature.G
gencodeFuncotation.setGeneTranscriptType( geneTranscriptType );
return this;
}

/**
* Set the Data Source Name {@link String} in the {@link GencodeFuncotation}.
* @param name The {@link String} containing the Data Source Name for the {@link GencodeFuncotation}.
* @return {@code this} {@link GencodeFuncotationBuilder}
*/
public GencodeFuncotationBuilder setDataSourceName( final String name ) {
gencodeFuncotation.setDataSourceName( name );
return this;
}
}
40 changes: 34 additions & 6 deletions ...dinstitute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationFactory.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -39,7 +39,12 @@
*/
public class GencodeFuncotationFactory extends DataSourceFuncotationFactory {

public static final String DATA_SOURCE_NAME = "Gencode";
//==================================================================================================================
// Public Static Members:
/**
* Default name for this data source.
*/
public static final String DEFAULT_NAME = "Gencode";

//==================================================================================================================
// Private Static Members:
Expand Down Expand Up @@ -117,6 +122,11 @@ public class GencodeFuncotationFactory extends DataSourceFuncotationFactory {
//==================================================================================================================
// Private Members:

/**
* The name of this Gencode data source.
*/
private final String name;

/**
* ReferenceSequenceFile for the transcript reference file.
*/
Expand Down Expand Up @@ -165,26 +175,27 @@ public class GencodeFuncotationFactory extends DataSourceFuncotationFactory {
// Constructors:

public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFile, final String version) {
this(gencodeTranscriptFastaFile, version, FuncotatorArgumentDefinitions.TRANSCRIPT_SELECTION_MODE_DEFAULT_VALUE, new HashSet<>(), new LinkedHashMap<>());
this(gencodeTranscriptFastaFile, version, DEFAULT_NAME, FuncotatorArgumentDefinitions.TRANSCRIPT_SELECTION_MODE_DEFAULT_VALUE, new HashSet<>(), new LinkedHashMap<>());
}

public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFile, final String version, final Set<String> userRequestedTranscripts) {
this(gencodeTranscriptFastaFile, version, FuncotatorArgumentDefinitions.TRANSCRIPT_SELECTION_MODE_DEFAULT_VALUE, userRequestedTranscripts, new LinkedHashMap<>());
this(gencodeTranscriptFastaFile, version, DEFAULT_NAME, FuncotatorArgumentDefinitions.TRANSCRIPT_SELECTION_MODE_DEFAULT_VALUE, userRequestedTranscripts, new LinkedHashMap<>());
}

public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFile, final String version,final TranscriptSelectionMode transcriptSelectionMode) {
this(gencodeTranscriptFastaFile, version, transcriptSelectionMode, new HashSet<>(), new LinkedHashMap<>());
this(gencodeTranscriptFastaFile, version, DEFAULT_NAME, transcriptSelectionMode, new HashSet<>(), new LinkedHashMap<>());
}

public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFile,
final String version,
final TranscriptSelectionMode transcriptSelectionMode,
final Set<String> userRequestedTranscripts) {
this(gencodeTranscriptFastaFile, version, transcriptSelectionMode, userRequestedTranscripts, new LinkedHashMap<>());
this(gencodeTranscriptFastaFile, version, DEFAULT_NAME, transcriptSelectionMode, userRequestedTranscripts, new LinkedHashMap<>());
}

public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFile,
final String version,
final String name,
final TranscriptSelectionMode transcriptSelectionMode,
final Set<String> userRequestedTranscripts,
final LinkedHashMap<String, String> annotationOverrides) {
Expand All @@ -198,6 +209,8 @@ public GencodeFuncotationFactory(final Path gencodeTranscriptFastaFile,

this.version = version;

this.name = name;

// Go through each requested transcript and remove the version numbers from them if they exist:
this.userRequestedTranscripts = new HashSet<>();
for ( final String transcript : userRequestedTranscripts ) {
Expand Down Expand Up @@ -231,7 +244,7 @@ public void close() {

@Override
public String getName() {
return DATA_SOURCE_NAME;
return name;
}

@Override
Expand Down Expand Up @@ -706,6 +719,9 @@ private GencodeFuncotation createCodingRegionFuncotationForNonProteinCodingFeatu
// Set the VariantClassification through a simple equivalency on the gene type (since we have no transcript info):
gencodeFuncotationBuilder.setVariantClassification( convertGeneTranscriptTypeToVariantClassification(exon.getGeneType()) );

// Set our data source name:
gencodeFuncotationBuilder.setDataSourceName(getName());

//==============================================================================================================

return gencodeFuncotationBuilder.build();
Expand Down Expand Up @@ -797,6 +813,9 @@ private GencodeFuncotation createCodingRegionFuncotationForProteinCodingFeature(
gencodeFuncotationBuilder.setVariantClassification( convertGeneTranscriptTypeToVariantClassification(exon.getGeneType()) );
}

// Set our data source name:
gencodeFuncotationBuilder.setDataSourceName(getName());

return gencodeFuncotationBuilder.build();
}

Expand Down Expand Up @@ -1127,6 +1146,9 @@ private GencodeFuncotation createUtrFuncotation(final VariantContext variant,
// Set our version:
gencodeFuncotationBuilder.setVersion(version);

// Set our data source name:
gencodeFuncotationBuilder.setDataSourceName(getName());

return gencodeFuncotationBuilder.build();
}

Expand Down Expand Up @@ -1203,6 +1225,9 @@ private GencodeFuncotation createIntronFuncotation(final VariantContext variant,
// Set our version:
gencodeFuncotationBuilder.setVersion(version);

// Set our data source name:
gencodeFuncotationBuilder.setDataSourceName(getName());

return gencodeFuncotationBuilder.build();
}

Expand Down Expand Up @@ -1800,6 +1825,9 @@ private GencodeFuncotation createIgrFuncotation(final VariantContext variant,
// Set our version:
funcotationBuilder.setVersion(version);

// Set our data source name:
funcotationBuilder.setDataSourceName(getName());

return funcotationBuilder.build();
}

Expand Down
1 change: 1 addition & 0 deletions ...main/java/org/broadinstitute/hellbender/tools/funcotator/mafOutput/MafOutputRenderer.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -494,6 +494,7 @@ protected void writeHeader(final LinkedHashMap<String, String> outputMap) {
printWriter.write(Funcotator.VERSION);
printWriter.write(" | Date ");
printWriter.write(new SimpleDateFormat("yyyymmdd'T'hhmmss").format(new Date()));
printWriter.write(" | ");
printWriter.write(getDataSourceInfoString());
writeLine("");

Expand Down
36 changes: 18 additions & 18 deletions src/test/java/org/broadinstitute/hellbender/tools/funcotator/FuncotatorIntegrationTest.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -203,24 +203,24 @@ public Object[][] provideForLargeDataValidationTest() {
true,
FuncotatorTestConstants.REFERENCE_VERSION_HG19,
},
{
"C828.TCGA-D3-A2JP-06A-11D-A19A-08.3-filtered.PASS.vcf",
"Homo_sapiens_assembly19.fasta",
true,
FuncotatorTestConstants.REFERENCE_VERSION_HG19
},
{
"hg38_test_variants.vcf",
"Homo_sapiens_assembly38.fasta",
false,
FuncotatorTestConstants.REFERENCE_VERSION_HG38
},
{
"sample21.trimmed.vcf",
"Homo_sapiens_assembly38.fasta",
false,
FuncotatorTestConstants.REFERENCE_VERSION_HG38
}
// {
// "C828.TCGA-D3-A2JP-06A-11D-A19A-08.3-filtered.PASS.vcf",
// "Homo_sapiens_assembly19.fasta",
// true,
// FuncotatorTestConstants.REFERENCE_VERSION_HG19
// },
// {
// "hg38_test_variants.vcf",
// "Homo_sapiens_assembly38.fasta",
// false,
// FuncotatorTestConstants.REFERENCE_VERSION_HG38
// },
// {
// "sample21.trimmed.vcf",
// "Homo_sapiens_assembly38.fasta",
// false,
// FuncotatorTestConstants.REFERENCE_VERSION_HG38
// }
};
}

Expand Down
3 changes: 2 additions & 1 deletion ...institute/hellbender/tools/funcotator/dataSources/gencode/GencodeFuncotationUnitTest.java
View file
Original file line number Diff line number Diff line change
Expand Up @@ -31,11 +31,12 @@ private static GencodeFuncotation createGencodeFuncotation(final String hugoSymb
final String cDnaChange, final String codonChange,
final String proteinChange, final Double gcContent,
final String referenceContext,
final List<String> otherTranscripts) {
final List<String> otherTranscripts) {

final GencodeFuncotation gencodeFuncotation = new GencodeFuncotation();

gencodeFuncotation.setVersion("TEST_VERSION");
gencodeFuncotation.setDataSourceName(GencodeFuncotationFactory.DEFAULT_NAME);

gencodeFuncotation.setHugoSymbol( hugoSymbol );
gencodeFuncotation.setNcbiBuild( ncbiBuild );
Expand Down
Loading

0 comments on commit 87b6ce8

Please sign in to comment.