public static Map<String, VCFHeader> getVCFHeadersFromRods(
GenomeAnalysisEngine toolkit, Collection<String> rodNames) {
Map<String, VCFHeader> data = new HashMap<String, VCFHeader>();
// iterate to get all of the sample names
List<ReferenceOrderedDataSource> dataSources = toolkit.getRodDataSources();
for (ReferenceOrderedDataSource source : dataSources) {
// ignore the rod if it's not in our list
if (rodNames != null && !rodNames.contains(source.getName())) continue;
if (source.getHeader() != null && source.getHeader() instanceof VCFHeader)
data.put(source.getName(), (VCFHeader) source.getHeader());
}
return data;
}
/**
* Gets the appropriately formatted header for a VCF file describing this GATK run
*
* @param engine the GATK engine that holds the walker name, GATK version, and other information
* @param argumentSources contains information on the argument values provided to the GATK for
* converting to a command line string. Should be provided from the data in the parsing
* engine. Can be empty in which case the command line will be the empty string.
* @return VCF header line describing this run of the GATK.
*/
public static VCFHeaderLine getCommandLineArgumentHeaderLine(
final GenomeAnalysisEngine engine, final Collection<Object> argumentSources) {
if (engine == null) throw new IllegalArgumentException("engine cannot be null");
if (argumentSources == null)
throw new IllegalArgumentException("argumentSources cannot be null");
final Map<String, String> attributes = new LinkedHashMap<>();
attributes.put("ID", engine.getWalkerName());
attributes.put("Version", CommandLineGATK.getVersionNumber());
final Date date = new Date();
attributes.put("Date", date.toString());
attributes.put("Epoch", Long.toString(date.getTime()));
attributes.put(
"CommandLineOptions",
engine.createApproximateCommandLineArgumentString(argumentSources.toArray()));
return new VCFSimpleHeaderLine(GATK_COMMAND_LINE_KEY, attributes);
}
/**
* Gets the header fields from all VCF rods input by the user
*
* @param toolkit GATK engine
* @param rodNames names of rods to use, or null if we should use all possible ones
* @return a set of all fields
*/
public static Set<VCFHeaderLine> getHeaderFields(
GenomeAnalysisEngine toolkit, Collection<String> rodNames) {
// keep a map of sample name to occurrences encountered
TreeSet<VCFHeaderLine> fields = new TreeSet<VCFHeaderLine>();
// iterate to get all of the sample names
List<ReferenceOrderedDataSource> dataSources = toolkit.getRodDataSources();
for (ReferenceOrderedDataSource source : dataSources) {
// ignore the rod if it's not in our list
if (rodNames != null && !rodNames.contains(source.getName())) continue;
if (source.getRecordType().equals(VariantContext.class)) {
VCFHeader header = (VCFHeader) source.getHeader();
if (header != null) fields.addAll(header.getMetaDataInSortedOrder());
}
}
return fields;
}