@Test(dataProvider = "variousFormatReaderTestCases")
public void variousFormatReaderTest(final String inputFile) throws IOException {
final File input = new File(TEST_DATA_DIR, inputFile);
final SamReader reader = SamReaderFactory.makeDefault().open(input);
for (final SAMRecord ignored : reader) {}
reader.close();
}
@Test(dataProvider = "composeAllPermutationsOfSamInputResource")
public void queryInputResourcePermutation(final SamInputResource resource) throws IOException {
final SamReader reader = SamReaderFactory.makeDefault().open(resource);
LOG.info(String.format("Query from %s ...", resource));
if (reader.hasIndex()) {
final StopWatch stopWatch = new StopWatch();
stopWatch.start();
final SAMRecordIterator q1 = reader.query("chr1", 500000, 100000000, true);
observedRecordOrdering1.add(Iterables.slurp(q1));
q1.close();
final SAMRecordIterator q20 = reader.query("chr20", 1, 1000000, true);
observedRecordOrdering20.add(Iterables.slurp(q20));
q20.close();
final SAMRecordIterator q3 = reader.query("chr3", 1, 10000000, true);
observedRecordOrdering3.add(Iterables.slurp(q3));
q3.close();
stopWatch.stop();
LOG.info(String.format("Finished queries in %sms", stopWatch.getElapsedTime()));
Assert.assertEquals(
observedRecordOrdering1.size(), 1, "read different records for chromosome 1");
Assert.assertEquals(
observedRecordOrdering20.size(), 1, "read different records for chromosome 20");
Assert.assertEquals(
observedRecordOrdering3.size(), 1, "read different records for chromosome 3");
} else if (resource.indexMaybe() != null) {
LOG.warn("Resource has an index source, but is not indexed: " + resource);
} else {
LOG.info("Skipping query operation: no index.");
}
reader.close();
}
@Test
public void testWrap() throws IOException {
final Path input = Paths.get(TEST_DATA_DIR.getPath(), "noheader.sam");
final SamReader wrappedReader =
SamReaderFactory.makeDefault().setPathWrapper(SamReaderFactoryTest::addHeader).open(input);
int records = countRecords(wrappedReader);
Assert.assertEquals(10, records);
}
@Test
public void testSamReaderFromURL() throws IOException {
final String samFilePath = new File(TEST_DATA_DIR, "unsorted.sam").getAbsolutePath();
final URL samURL = new URL("file://" + samFilePath);
final SamReaderFactory factory =
SamReaderFactory.makeDefault().validationStringency(ValidationStringency.SILENT);
final SamReader reader = factory.open(SamInputResource.of(samURL));
Assert.assertEquals(countRecords(reader), 10);
}
@Test
public void testSamReaderFromSeekableStream() throws IOException {
// even though a SAM isn't indexable, make sure we can open one
// using a seekable stream
final File samFile = new File(TEST_DATA_DIR, "unsorted.sam");
final SamReaderFactory factory =
SamReaderFactory.makeDefault().validationStringency(ValidationStringency.SILENT);
final SamReader reader = factory.open(SamInputResource.of(new SeekableFileStream(samFile)));
Assert.assertEquals(countRecords(reader), 10);
}
@Test
public void openPath() throws IOException {
final Path path = localBam.toPath();
final List<SAMRecord> records;
final SAMFileHeader fileHeader;
try (final SamReader reader = SamReaderFactory.makeDefault().open(path)) {
LOG.info(String.format("Reading from %s ...", path));
records = Iterables.slurp(reader);
fileHeader = reader.getFileHeader();
reader.close();
}
try (final SamReader fileReader = SamReaderFactory.makeDefault().open(localBam)) {
final List<SAMRecord> expectedRecords = Iterables.slurp(fileReader);
final SAMFileHeader expectedFileHeader = fileReader.getFileHeader();
Assert.assertEquals(records, expectedRecords);
Assert.assertEquals(fileHeader, expectedFileHeader);
}
}
@Test(expectedExceptions = SAMFormatException.class)
public void testSamReaderFromMalformedSeekableStream() throws IOException {
// use a bogus (.bai file) to force SamReaderFactory to fall through to the
// fallback code that assumes a SAM File when it can't determine the
// format of the input, to ensure that it results in a SAMFormatException
final File samFile = new File(TEST_DATA_DIR, "cram_with_bai_index.cram.bai");
final SamReaderFactory factory =
SamReaderFactory.makeDefault().validationStringency(ValidationStringency.SILENT);
final SamReader reader = factory.open(SamInputResource.of(new SeekableFileStream(samFile)));
countRecords(reader);
}
@Test
public void checkHasIndexForStreamingPathBamWithFileIndex() throws IOException {
InputResource bam = new NeverFilePathInputResource(localBam.toPath());
InputResource index = new FileInputResource(localBamIndex);
// ensure that the index is being used, not checked in queryInputResourcePermutation
try (final SamReader reader =
SamReaderFactory.makeDefault().open(new SamInputResource(bam, index))) {
Assert.assertTrue(reader.hasIndex());
}
}
// See https://github.com/samtools/htsjdk/issues/76
@Test(dataProvider = "queryIntervalIssue76TestCases")
public void queryIntervalIssue76(
final String sequenceName, final int start, final int end, final int expectedCount)
throws IOException {
final File input = new File(TEST_DATA_DIR, "issue76.bam");
final SamReader reader = SamReaderFactory.makeDefault().open(input);
final QueryInterval interval =
new QueryInterval(
reader.getFileHeader().getSequence(sequenceName).getSequenceIndex(), start, end);
Assert.assertEquals(countRecordsInQueryInterval(reader, interval), expectedCount);
reader.close();
}
@Test(dataProvider = "composeAllPermutationsOfSamInputResource")
public void exhaustInputResourcePermutation(final SamInputResource resource) throws IOException {
final SamReader reader = SamReaderFactory.makeDefault().open(resource);
LOG.info(String.format("Reading from %s ...", resource));
final List<SAMRecord> slurped = Iterables.slurp(reader);
final SAMFileHeader fileHeader = reader.getFileHeader();
reader.hasIndex();
reader.indexing().hasBrowseableIndex();
reader.close();
/* Ensure all tests have read the same records in the same order or, if this is the first test, set it as the template. */
observedHeaders.add(fileHeader);
observedRecordOrdering.add(slurped);
Assert.assertEquals(observedHeaders.size(), 1, "read different headers than other testcases");
Assert.assertEquals(
observedRecordOrdering.size(), 1, "read different records than other testcases");
}
@Test(dataProvider = "variousFormatReaderTestCases")
public void samRecordFactoryTest(final String inputFile) throws IOException {
final File input = new File(TEST_DATA_DIR, inputFile);
final SAMRecordFactoryTester recordFactory = new SAMRecordFactoryTester();
final SamReaderFactory readerFactory =
SamReaderFactory.makeDefault().samRecordFactory(recordFactory);
final SamReader reader = readerFactory.open(input);
int i = 0;
for (final SAMRecord ignored : reader) {
++i;
}
reader.close();
Assert.assertTrue(i > 0);
if (inputFile.endsWith(".sam") || inputFile.endsWith(".sam.gz"))
Assert.assertEquals(recordFactory.samRecordsCreated, i);
else if (inputFile.endsWith(".bam")) Assert.assertEquals(recordFactory.bamRecordsCreated, i);
}
@Test
public void customReaderFactoryTest() throws IOException {
try {
CustomReaderFactory.setInstance(
new CustomReaderFactory(
"https://www.googleapis.com/genomics/v1beta/reads/,"
+ "htsjdk.samtools.SamReaderFactoryTest$TestReaderFactory"));
final SamReader reader =
SamReaderFactory.makeDefault()
.open(
SamInputResource.of(
"https://www.googleapis.com/genomics/v1beta/reads/?uncompressed.sam"));
int i = 0;
for (@SuppressWarnings("unused") final SAMRecord ignored : reader) {
++i;
}
reader.close();
Assert.assertTrue(i > 0);
} finally {
CustomReaderFactory.resetToDefaultInstance();
}
}
private void getCRAMReaderFromInputResource(
final BiFunction<URL, URL, SamInputResource> getInputResource,
final boolean hasIndex,
final int expectedCount)
throws IOException {
final String cramFilePath =
new File(TEST_DATA_DIR, "cram_with_bai_index.cram").getAbsolutePath();
final String cramIndexPath =
new File(TEST_DATA_DIR, "cram_with_bai_index.cram.bai").getAbsolutePath();
final URL cramURL = new URL("file://" + cramFilePath);
final URL indexURL = new URL("file://" + cramIndexPath);
final SamReaderFactory factory =
SamReaderFactory.makeDefault()
.referenceSource(new ReferenceSource(new File(TEST_DATA_DIR, "hg19mini.fasta")))
.validationStringency(ValidationStringency.SILENT);
final SamReader reader = factory.open(getInputResource.apply(cramURL, indexURL));
int count =
hasIndex
? countRecordsInQueryInterval(reader, new QueryInterval(1, 10, 1000))
: countRecords(reader);
Assert.assertEquals(count, expectedCount);
}
@Override
public SamReader open(URL url) {
final File file = new File(TEST_DATA_DIR, url.getQuery());
LOG.info("Opening customr reader for " + file.toString());
return SamReaderFactory.makeDefault().open(file);
}
