@Test(dataProvider = "data")
public void testProgramGroupAndReadGroupMerge(File inputFiles[], File expectedOutputFile)
throws IOException {
BufferedReader reader = new BufferedReader(new FileReader(expectedOutputFile));
String line;
String expected_output = "";
while ((line = reader.readLine()) != null) {
expected_output += line + "\n";
}
final List<SAMFileReader> readers = new ArrayList<SAMFileReader>();
final List<SAMFileHeader> headers = new ArrayList<SAMFileHeader>();
for (final File inFile : inputFiles) {
IOUtil.assertFileIsReadable(inFile);
final SAMFileReader in = new SAMFileReader(inFile);
// We are now checking for zero-length reads, so suppress complaint about that.
in.setValidationStringency(ValidationStringency.SILENT);
readers.add(in);
headers.add(in.getFileHeader());
}
final MergingSamRecordIterator iterator;
final SamFileHeaderMerger headerMerger =
new SamFileHeaderMerger(SAMFileHeader.SortOrder.coordinate, headers, true);
iterator = new MergingSamRecordIterator(headerMerger, readers, false);
ByteArrayOutputStream baos = new ByteArrayOutputStream();
SAMFileWriter writer =
new SAMFileWriterFactory().makeSAMWriter(headerMerger.getMergedHeader(), true, baos);
while (iterator.hasNext()) {
writer.addAlignment(iterator.next());
}
writer.close();
String actual_output = StringUtil.bytesToString(baos.toByteArray());
List<String> actual = Arrays.asList(actual_output.split("\\n"));
List<String> expected = Arrays.asList(expected_output.split("\\n"));
for (int i = 0; i < expected.size(); i++) {
if (expected.get(i).startsWith("@")) {
Assert.assertTrue(headersEquivalent(actual.get(i), expected.get(i)));
} else {
List<String> expectedSamParts = Arrays.asList(expected.get(i).split("\\s*"));
List<String> actualSamParts = Arrays.asList(actual.get(i).split("\\s*"));
for (String exp : expectedSamParts) {
Assert.assertTrue(actualSamParts.contains(exp));
}
for (String act : actualSamParts) {
Assert.assertTrue(expectedSamParts.contains(act));
}
}
}
}
/** Tests that we can successfully merge two files with */
@Test
public void testMerging() {
File INPUT[] = {
new File(TEST_DATA_DIR, "SamFileHeaderMergerTest/Chromosome1to10.bam"),
new File(TEST_DATA_DIR, "SamFileHeaderMergerTest/Chromosome5to9.bam")
};
final List<SAMFileReader> readers = new ArrayList<SAMFileReader>();
final List<SAMFileHeader> headers = new ArrayList<SAMFileHeader>();
for (final File inFile : INPUT) {
IOUtil.assertFileIsReadable(inFile);
final SAMFileReader in = new SAMFileReader(inFile);
// We are now checking for zero-length reads, so suppress complaint about that.
in.setValidationStringency(ValidationStringency.SILENT);
readers.add(in);
headers.add(in.getFileHeader());
}
final MergingSamRecordIterator iterator;
final SamFileHeaderMerger headerMerger =
new SamFileHeaderMerger(SAMFileHeader.SortOrder.unsorted, headers, true);
iterator = new MergingSamRecordIterator(headerMerger, readers, false);
headerMerger.getMergedHeader();
// count the total reads, and record read counts for each sequence
Map<Integer, Integer> seqCounts = new HashMap<Integer, Integer>();
int totalCount = 0;
while (iterator.hasNext()) {
SAMRecord r = iterator.next();
if (seqCounts.containsKey(r.getReferenceIndex())) {
seqCounts.put(r.getReferenceIndex(), seqCounts.get(r.getReferenceIndex()) + 1);
} else {
seqCounts.put(r.getReferenceIndex(), 1);
}
++totalCount;
}
assertEquals(totalCount, 1500);
for (Integer i : seqCounts.keySet()) {
if (i < 4 || i > 8) {
// seqeunce 5 - 9 should have 200 reads (indices 4 - 8)
assertEquals(seqCounts.get(i).intValue(), 100);
} else {
// the others should have 100
assertEquals(seqCounts.get(i).intValue(), 200);
}
}
}