@Test
public void testCountsFromLocusTraversal() {
final GenomeAnalysisEngine engine = new GenomeAnalysisEngine();
engine.setGenomeLocParser(genomeLocParser);
final Collection<SAMReaderID> samFiles = new ArrayList<>();
final SAMReaderID readerID = new SAMReaderID(testBAM, new Tags());
samFiles.add(readerID);
final SAMDataSource dataSource =
new SAMDataSource(
samFiles,
new ThreadAllocation(),
null,
genomeLocParser,
false,
SAMFileReader.ValidationStringency.STRICT,
null,
null,
new ValidationExclusion(),
new ArrayList<ReadFilter>(),
new ArrayList<ReadTransformer>(),
false,
(byte) 30,
false,
true);
engine.setReadsDataSource(dataSource);
final Set<String> samples = SampleUtils.getSAMFileSamples(dataSource.getHeader());
final TraverseLociNano traverseLociNano = new TraverseLociNano(1);
final DummyLocusWalker walker = new DummyLocusWalker();
traverseLociNano.initialize(engine, walker, null);
for (final Shard shard : dataSource.createShardIteratorOverAllReads(new LocusShardBalancer())) {
final WindowMaker windowMaker =
new WindowMaker(
shard, genomeLocParser, dataSource.seek(shard), shard.getGenomeLocs(), samples);
for (WindowMaker.WindowMakerIterator window : windowMaker) {
final LocusShardDataProvider dataProvider =
new LocusShardDataProvider(
shard,
shard.getReadProperties(),
genomeLocParser,
window.getLocus(),
window,
reference,
new ArrayList<ReferenceOrderedDataSource>());
traverseLociNano.traverse(walker, dataProvider, 0);
dataProvider.close();
}
windowMaker.close();
}
// dataSource.close();
Assert.assertEquals(
engine.getCumulativeMetrics().getNumReadsSeen(), contigs.size() * numReadsPerContig);
Assert.assertEquals(
engine.getCumulativeMetrics().getNumIterations(), contigs.size() * numReadsPerContig);
}
@Test
public void testFilteredCounts() {
final GenomeAnalysisEngine engine = new GenomeAnalysisEngine();
engine.setGenomeLocParser(genomeLocParser);
final Collection<SAMReaderID> samFiles = new ArrayList<>();
final SAMReaderID readerID = new SAMReaderID(testBAM, new Tags());
samFiles.add(readerID);
final List<ReadFilter> filters = new ArrayList<>();
filters.add(new EveryTenthReadFilter());
final SAMDataSource dataSource =
new SAMDataSource(
samFiles,
new ThreadAllocation(),
null,
genomeLocParser,
false,
SAMFileReader.ValidationStringency.STRICT,
null,
null,
new ValidationExclusion(),
filters,
new ArrayList<ReadTransformer>(),
false,
(byte) 30,
false,
true);
engine.setReadsDataSource(dataSource);
final TraverseReadsNano traverseReadsNano = new TraverseReadsNano(1);
final DummyReadWalker walker = new DummyReadWalker();
traverseReadsNano.initialize(engine, walker, null);
for (final Shard shard : dataSource.createShardIteratorOverAllReads(new ReadShardBalancer())) {
final ReadShardDataProvider dataProvider =
new ReadShardDataProvider(
shard,
engine.getGenomeLocParser(),
dataSource.seek(shard),
reference,
new ArrayList<ReferenceOrderedDataSource>());
traverseReadsNano.traverse(walker, dataProvider, 0);
dataProvider.close();
}
Assert.assertEquals(
(long)
engine
.getCumulativeMetrics()
.getCountsByFilter()
.get(EveryTenthReadFilter.class.getSimpleName()),
contigs.size() * numReadsPerContig / 10);
}
/**
* this is the function that the inheriting class can expect to have called when the command line
* system has initialized.
*
* @return the return code to exit the program with
*/
protected int execute() throws Exception {
engine.setParser(parser);
argumentSources.add(this);
Walker<?, ?> walker = engine.getWalkerByName(getAnalysisName());
try {
// Make sure a valid GATK user key is present, if required.
authorizeGATKRun();
engine.setArguments(getArgumentCollection());
// File lists can require a bit of additional expansion. Set these explicitly by the engine.
engine.setSAMFileIDs(
ListFileUtils.unpackBAMFileList(getArgumentCollection().samFiles, parser));
engine.setWalker(walker);
walker.setToolkit(engine);
Collection<ReadFilter> filters = engine.createFilters();
engine.setFilters(filters);
// load the arguments into the walker / filters.
// TODO: The fact that this extra load call exists here when all the parsing happens at the
// engine
// TODO: level indicates that we're doing something wrong. Turn this around so that the GATK
// can drive
// TODO: argument processing.
loadArgumentsIntoObject(walker);
argumentSources.add(walker);
Collection<RMDTriplet> rodBindings =
ListFileUtils.unpackRODBindings(parser.getRodBindings(), parser);
engine.setReferenceMetaDataFiles(rodBindings);
for (ReadFilter filter : filters) {
loadArgumentsIntoObject(filter);
argumentSources.add(filter);
}
engine.execute();
generateGATKRunReport(walker);
} catch (Exception e) {
generateGATKRunReport(walker, e);
throw e;
}
// always return 0
return 0;
}
/**
* GATK provides the walker as an argument source.
*
* @return List of walkers to load dynamically.
*/
@Override
protected Class[] getArgumentSources() {
// No walker info? No plugins.
if (getAnalysisName() == null) return new Class[] {};
Collection<Class> argumentSources = new ArrayList<Class>();
Walker walker = engine.getWalkerByName(getAnalysisName());
engine.setArguments(getArgumentCollection());
engine.setWalker(walker);
walker.setToolkit(engine);
argumentSources.add(walker.getClass());
Collection<ReadFilter> filters = engine.createFilters();
for (ReadFilter filter : filters) argumentSources.add(filter.getClass());
Class[] argumentSourcesAsArray = new Class[argumentSources.size()];
return argumentSources.toArray(argumentSourcesAsArray);
}
@Override
protected String getArgumentSourceName(Class argumentSource) {
return engine.getWalkerName((Class<Walker>) argumentSource);
}