@Test
public void inputResourceFromStringTest() throws IOException {
Assert.assertEquals(
SamInputResource.of("http://test.url").data().type(), InputResource.Type.URL);
Assert.assertEquals(
SamInputResource.of("https://test.url").data().type(), InputResource.Type.URL);
Assert.assertEquals(
SamInputResource.of("ftp://test.url").data().type(), InputResource.Type.URL);
Assert.assertEquals(SamInputResource.of("/a/b/c").data().type(), InputResource.Type.FILE);
}
@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 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(expectedExceptions = RuntimeIOException.class)
public void testCRAMReaderFromURLBadIndexFile() throws IOException {
// deliberately specify a bad index file to ensure we get an IOException
getCRAMReaderFromInputResource(
(cramURL, indexURL) -> {
return SamInputResource.of(cramURL).index(new File("nonexistent.bai"));
},
true,
3);
}
@Test
public void testCRAMReaderFromURLNoIndexFile() throws IOException {
// get just a CRAM reader (no index) from an URL-backed resource
getCRAMReaderFromInputResource(
(cramURL, indexURL) -> {
return SamInputResource.of(cramURL);
},
false,
11);
}
@Test
public void testCRAMReaderFromURL() throws IOException {
// get a CRAM reader with an index from a URL-backed resource
getCRAMReaderFromInputResource(
(cramURL, indexURL) -> {
return SamInputResource.of(cramURL).index(indexURL);
},
true,
3);
}
@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);
}
private void runLocalRemoteTest(
final URL bamURL,
final File bamFile,
final String sequence,
final int startPos,
final int endPos,
final boolean contained) {
verbose("Testing query " + sequence + ":" + startPos + "-" + endPos + " ...");
final SamReader reader1 =
SamReaderFactory.makeDefault()
.disable(SamReaderFactory.Option.EAGERLY_DECODE)
.open(SamInputResource.of(bamFile).index(BAM_INDEX_FILE));
final SamReader reader2 =
SamReaderFactory.makeDefault()
.disable(SamReaderFactory.Option.EAGERLY_DECODE)
.open(SamInputResource.of(bamURL).index(BAM_INDEX_FILE));
final Iterator<SAMRecord> iter1 = reader1.query(sequence, startPos, endPos, contained);
final Iterator<SAMRecord> iter2 = reader2.query(sequence, startPos, endPos, contained);
final List<SAMRecord> records1 = new ArrayList<SAMRecord>();
final List<SAMRecord> records2 = new ArrayList<SAMRecord>();
while (iter1.hasNext()) {
records1.add(iter1.next());
}
while (iter2.hasNext()) {
records2.add(iter2.next());
}
assertTrue(records1.size() > 0);
assertEquals(records1.size(), records2.size());
for (int i = 0; i < records1.size(); i++) {
// System.out.println(records1.get(i).format());
assertEquals(records1.get(i).getSAMString(), records2.get(i).getSAMString());
}
}
private List<String> getReferenceNames(final URL bamFile) throws IOException {
final SamReader reader =
SamReaderFactory.makeDefault().open(SamInputResource.of(bamFile.openStream()));
final List<String> result = new ArrayList<String>();
final List<SAMSequenceRecord> seqRecords =
reader.getFileHeader().getSequenceDictionary().getSequences();
for (final SAMSequenceRecord seqRecord : seqRecords) {
if (seqRecord.getSequenceName() != null) {
result.add(seqRecord.getSequenceName());
}
}
reader.close();
return result;
}
@Test
public void testCRAMReaderFromURLStream() throws IOException {
// get a CRAM reader with an index from a stream-backed resource created from a URL
getCRAMReaderFromInputResource(
(cramURL, indexURL) -> {
try {
ISeekableStreamFactory streamFactory = SeekableStreamFactory.getInstance();
return SamInputResource.of(streamFactory.getStreamFor(cramURL))
.index(streamFactory.getStreamFor(indexURL));
} catch (IOException e) {
throw new RuntimeIOException(e);
}
},
true,
3);
}
public SamReader getSamReader(ResourceLocator locator, boolean requireIndex) throws IOException {
if (requireIndex) {
final SamReaderFactory factory =
SamReaderFactory.makeDefault().validationStringency(ValidationStringency.SILENT);
SeekableStream indexStream = getIndexStream(locator.getBamIndexPath());
this.indexed = true;
SeekableStream ss =
new IGVSeekableBufferedStream(
IGVSeekableStreamFactory.getInstance().getStreamFor(url), 128000);
SamInputResource resource = SamInputResource.of(ss).index(indexStream);
return factory.open(resource);
} else {
InputStream is = HttpUtils.getInstance().openConnectionStream(url);
return new SAMFileReader(new BufferedInputStream(is));
}
}
@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 int runQueryTest(
final URL bamURL,
final String sequence,
final int startPos,
final int endPos,
final boolean contained) {
verbose("Testing query " + sequence + ":" + startPos + "-" + endPos + " ...");
final SamReader reader1 =
SamReaderFactory.makeDefault()
.disable(SamReaderFactory.Option.EAGERLY_DECODE)
.open(SamInputResource.of(bamURL).index(BAM_INDEX_FILE));
final SamReader reader2 =
SamReaderFactory.makeDefault()
.disable(SamReaderFactory.Option.EAGERLY_DECODE)
.open(SamInputResource.of(bamURL).index(BAM_INDEX_FILE));
final Iterator<SAMRecord> iter1 = reader1.query(sequence, startPos, endPos, contained);
final Iterator<SAMRecord> iter2 = reader2.iterator();
// Compare ordered iterators.
// Confirm that iter1 is a subset of iter2 that properly filters.
SAMRecord record1 = null;
SAMRecord record2 = null;
int count1 = 0;
int count2 = 0;
int beforeCount = 0;
int afterCount = 0;
while (true) {
if (record1 == null && iter1.hasNext()) {
record1 = iter1.next();
count1++;
}
if (record2 == null && iter2.hasNext()) {
record2 = iter2.next();
count2++;
}
// System.out.println("Iteration:");
// System.out.println(" Record1 = " + ((record1 == null) ? "null" : record1.format()));
// System.out.println(" Record2 = " + ((record2 == null) ? "null" : record2.format()));
if (record1 == null && record2 == null) {
break;
}
if (record1 == null) {
checkPassesFilter(false, record2, sequence, startPos, endPos, contained);
record2 = null;
afterCount++;
continue;
}
assertNotNull(record2);
final int ordering = compareCoordinates(record1, record2);
if (ordering > 0) {
checkPassesFilter(false, record2, sequence, startPos, endPos, contained);
record2 = null;
beforeCount++;
continue;
}
assertTrue(ordering == 0);
checkPassesFilter(true, record1, sequence, startPos, endPos, contained);
checkPassesFilter(true, record2, sequence, startPos, endPos, contained);
assertEquals(record1.getReadName(), record2.getReadName());
assertEquals(record1.getReadString(), record2.getReadString());
record1 = null;
record2 = null;
}
CloserUtil.close(reader1);
CloserUtil.close(reader2);
verbose("Checked " + count1 + " records against " + count2 + " records.");
verbose("Found " + (count2 - beforeCount - afterCount) + " records matching.");
verbose("Found " + beforeCount + " records before.");
verbose("Found " + afterCount + " records after.");
return count1;
}
