@Override
public QueryFuture<FeatureSet> getQuery(FeatureSet set, int queryNum) {
if (queryNum == 0) {
/// limits us to CHROM #21
return SWQEFactory.getQueryInterface()
.getFeaturesByAttributes(
0,
set,
new RPNStack(
new RPNStack.FeatureAttribute("seqid"),
new RPNStack.Constant("21"),
RPNStack.Operation.EQUAL));
} else if (queryNum == 1) {
// limits us to the range of 20000000 through 30000000
return SWQEFactory.getQueryInterface()
.getFeaturesByRange(0, set, QueryInterface.Location.INCLUDES, "21", 20000000, 30000000);
} else {
// limits us to features with a particular tag
return SWQEFactory.getQueryInterface()
.getFeaturesByAttributes(
0,
set,
new RPNStack(
new RPNStack.TagOccurrence(
QueryVCFDumperTest.AD_HOC_TAG_SET, "non_synonymous_codon")));
}
}
/** export. */
public void export() {
if (args.length < 1 || args.length > 2) {
System.err.println(args.length + " arguments found");
System.out.println("VCFDumper <featureSetID> [outputFile]");
System.exit(-1);
}
// parse a SGID from a String representation, we need a more elegant solution here
String featureSetID = args[0];
SGID sgid = Utility.parseSGID(featureSetID);
FeatureSet fSet = SWQEFactory.getQueryInterface().getLatestAtomBySGID(sgid, FeatureSet.class);
// if this featureSet does not exist
if (fSet == null) {
System.out.println("featureSet ID not found");
System.exit(-2);
}
dumpVCFFromFeatureSetID(fSet, (args.length == 2 ? args[1] : null));
}
/**
* dumpVCFFromFeatureSetID.
*
* @param fSet a {@link com.github.seqware.queryengine.model.FeatureSet} object.
* @param file a {@link java.lang.String} object.
*/
public static void dumpVCFFromFeatureSetID(FeatureSet fSet, String file) {
BufferedWriter outputStream = null;
try {
if (file != null) {
outputStream = new BufferedWriter(new FileWriter(file));
} else {
outputStream = new BufferedWriter(new OutputStreamWriter(System.out));
}
outputStream.append("#CHROM POS\tID\tREF\tALT\tQUAL\tFILTER\tINFO\n");
} catch (IOException e) {
Logger.getLogger(VCFDumper.class.getName())
.fatal("Exception thrown starting export to file:", e);
System.exit(-1);
}
boolean caughtNonVCF = false;
boolean mrSuccess = false;
if (SWQEFactory.getQueryInterface() instanceof MRHBasePersistentBackEnd) {
// hack to use VCF MR
if (SWQEFactory.getModelManager() instanceof MRHBaseModelManager) {
try {
// pretend that the included
// com.github.seqware.queryengine.plugins.hbasemr.MRFeaturesByAttributesPlugin is an
// external plug-in
Class<? extends PluginInterface> arbitraryPlugin;
arbitraryPlugin = VCFDumperPlugin.class;
// get a FeatureSet from the back-end
QueryFuture<File> future =
SWQEFactory.getQueryInterface().getFeaturesByPlugin(0, arbitraryPlugin, fSet);
File get = future.get();
BufferedReader in = new BufferedReader(new FileReader(get));
IOUtils.copy(in, outputStream);
in.close();
get.deleteOnExit();
outputStream.flush();
outputStream.close();
mrSuccess = true;
} catch (IOException e) {
// fail out on IO error
Logger.getLogger(VCFDumper.class.getName())
.fatal("Exception thrown exporting to file:", e);
System.exit(-1);
} catch (Exception e) {
Logger.getLogger(VCFDumper.class.getName())
.info("MapReduce exporting failed, falling-through to normal exporting to file");
// fall-through and do normal exporting if Map Reduce exporting fails
}
} // TODO: clearly this should be expanded to include closing database etc
}
if (mrSuccess) {
return;
}
// fall-through if plugin-fails
try {
for (Feature feature : fSet) {
StringBuffer buffer = new StringBuffer();
boolean caught = outputFeatureInVCF(buffer, feature);
if (caught) {
caughtNonVCF = true;
}
outputStream.append(buffer);
outputStream.newLine();
}
outputStream.flush();
} catch (IOException e) {
Logger.getLogger(VCFDumper.class.getName()).fatal("Exception thrown exporting to file:", e);
System.exit(-1);
} finally {
IOUtils.closeQuietly(outputStream);
}
}