public void testTaxonomySDF() throws IOException {
try (final TestDirectory dir = new TestDirectory("sdfstats")) {
// One without tax
final File smallSdf = ReaderTestUtils.getDNADir(dir);
final String err =
checkMainInitBadFlags("-i", smallSdf.getPath(), "-o", "-", "--taxons", "41431");
TestUtils.containsAll(err, "does not contain taxonomy");
// One with tax
final File fullSdf = new File(dir, "sdf_full");
final File faOut = new File(dir, "ex.fasta.gz");
makeTestTaxonSdf(fullSdf);
checkMainInitOk("-i", fullSdf.getPath(), "-o", faOut.getPath(), "--taxons", "41431");
final String faout = FileHelper.gzFileToString(faOut);
TestUtils.containsAll(
faout,
"gi|218169684|gb|CP001289.1|",
"gi|218165370|gb|CP001287.1|",
"gi|218169729|gb|CP001290.1|",
"gi|218169631|gb|CP001288.1|");
}
}
public void testFullName() throws IOException {
try (TestDirectory dir = new TestDirectory("sdf2fasta")) {
final File sdf = ReaderTestUtils.getDNADir(FULL_NAME_DATA, new File(dir, "sdf"));
final File fasta = new File(dir, "fs.fasta.gz");
checkMainInitOk("-i", sdf.getPath(), "-o", fasta.getPath());
assertEquals(FULL_NAME_DATA, FileHelper.gzFileToString(fasta));
}
}
public void testEmptyFasta() throws IOException {
final SequencesReader reader = ReaderTestUtils.getReaderDNA("", mDir, new SdfId(0L));
reader.close();
final CompressedMemorySequencesReader cmsr =
new CompressedMemorySequencesReader(
mDir, new IndexFile(mDir), 5, true, false, new LongRange(0, 0));
assertEquals(0, cmsr.numberSequences());
}
@Override
public void testEquals() throws IOException {
final File dir = FileHelper.createTempDirectory(mDir);
final File otherDir = FileHelper.createTempDirectory(mDir);
ReaderTestUtils.getReaderDNAFastq("", dir, false).close();
ReaderTestUtils.getReaderDNAFastq("", otherDir, false).close();
final CompressedMemorySequencesReader cmsr =
new CompressedMemorySequencesReader(
dir, new IndexFile(dir), 5, true, false, new LongRange(0, 0));
final CompressedMemorySequencesReader other =
new CompressedMemorySequencesReader(
otherDir, new IndexFile(dir), 5, true, false, new LongRange(0, 0));
assertTrue(cmsr.equals(cmsr));
assertFalse(cmsr.equals(null));
assertFalse(cmsr.equals(other));
assertFalse(cmsr.equals("FDSA"));
}
public void testOverEndFastq() throws Exception {
final SequencesReader reader = ReaderTestUtils.getReaderDNAFastq(POSITIONS_FASTQ, mDir, false);
reader.close();
new CompressedMemorySequencesReader(
mDir, new IndexFile(mDir), 5, true, false, new LongRange(3, 8));
// Now this is OK rather than Exception, since our std behaviour is to warn and clip end that
// are too high back to the available num seqs
}
public static void makeTestTaxonSdf(File dest) throws IOException {
final String sequences =
FileHelper.resourceToString("com/rtg/reader/resources/sequences.fasta");
ReaderTestUtils.getReaderDNA(sequences, dest, null).close();
// cp taxonomy files to sdf
final File taxonomy = new File(dest, "taxonomy.tsv");
FileHelper.resourceToFile("com/rtg/reader/resources/taxonomy.tsv", taxonomy);
final File taxonomyLookup = new File(dest, "taxonomy_lookup.tsv");
FileHelper.resourceToFile("com/rtg/reader/resources/taxonomy_lookup.tsv", taxonomyLookup);
}
public void testFastqEnd() throws Exception {
final SequencesReader reader =
ReaderTestUtils.getReaderDNAFastq(POSITIONS_FASTQ, mDir, 20, false);
reader.close();
final SequencesIterator cmsr =
new CompressedMemorySequencesReader(
mDir, new IndexFile(mDir), 5, true, false, new LongRange(5, 6))
.iterator();
cmsr.nextSequence();
checkFastq(cmsr, "r5", "ACGTAC");
assertFalse(cmsr.nextSequence());
}
public void testRanges() throws Exception {
final SequencesReader reader = ReaderTestUtils.getReaderDNAFastq(SMALL_FASTQ, mDir, 20, false);
reader.close();
final CompressedMemorySequencesReader cmsr =
new CompressedMemorySequencesReader(
mDir, new IndexFile(mDir), 5, true, false, LongRange.NONE);
final byte[] foo = new byte[6];
cmsr.readQuality(0, foo, 4, 6);
final String qual = "EFGHIJ";
for (int i = 0; i < foo.length; i++) {
assertEquals(qual.charAt(i) - '!', foo[i]);
}
}
