/**
* This is the constructor that code should typically call when a SequencesReader is already
* available.
*
* @param genome SDF containing sequences and reference specification file.
* @param sex for this instance.
* @param fallback create a default ReferenceGenome of the given type when no reference file is
* present, if null and there is no reference an error will be thrown
* @throws IOException when actual I/O error or problems in file definition.
*/
public ReferenceGenome(final SequencesReader genome, final Sex sex, DefaultFallback fallback)
throws IOException {
try (BufferedReader r = getReferenceReader(genome, fallback)) {
parse(ReaderUtils.getSequenceLengthMap(genome), r, sex);
}
}
/**
* Constructor for testing.
*
* @param genome with names of sequences.
* @param reference contents of description of reference genome.
* @param sex for this instance.
* @throws IOException when actual I/O error or problems in file definition.
*/
public ReferenceGenome(final SequencesReader genome, final Reader reference, final Sex sex)
throws IOException {
parse(ReaderUtils.getSequenceLengthMap(genome), reference, sex);
}
public void testExclusionCases() throws IOException {
final StringReader reader = new StringReader(PSEUDO_REF_2);
final MockSequencesReader genome = new MockSequencesReader(SequenceType.DNA, 3, 300000);
final ReferenceGenome rg = new ReferenceGenome(genome, reader, Sex.MALE);
final Map<String, Long> sequenceNameMap = ReaderUtils.getSequenceNameMap(genome);
// front overlap
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {new HashingRegion(1, 5000, 1, 59999, 4500, 59999)},
new HashingRegion[] {new HashingRegion(1, 5000, 1, 65000, 4500, 65500)});
// rear overlap
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {new HashingRegion(1, 80000, 1, 90000, 80000, 90500)},
new HashingRegion[] {new HashingRegion(1, 70000, 1, 90000, 69500, 90500)});
// region entirely within PAR region
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {},
new HashingRegion[] {new HashingRegion(1, 70000, 1, 75000, 69500, 75500)});
// region entirely within PAR region
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {},
new HashingRegion[] {new HashingRegion(1, 60000, 1, 80000, 59500, 80500)});
// region entirely within PAR region
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {},
new HashingRegion[] {new HashingRegion(1, 59999, 1, 80000, 59500, 80500)});
// region entirely consumes PAR region
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {
new HashingRegion(1, 55000, 1, 59999, 54500, 59999),
new HashingRegion(1, 80000, 1, 85000, 80000, 85500)
},
new HashingRegion[] {new HashingRegion(1, 55000, 1, 85000, 54500, 85500)});
// region on edges
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {
new HashingRegion(1, 55000, 1, 59999, 54500, 59999),
new HashingRegion(1, 80000, 1, 85000, 80000, 85500)
},
new HashingRegion[] {
new HashingRegion(1, 55000, 1, 60000, 54500, 60500),
new HashingRegion(1, 60000, 1, 80000, 59500, 80500),
new HashingRegion(1, 80000, 1, 85000, 79500, 85500)
});
// front overlap padding
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {new HashingRegion(1, 5000, 1, 59999, 4500, 59999)},
new HashingRegion[] {new HashingRegion(1, 5000, 1, 65000, 4500, 70000)});
// rear overlap padding
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {new HashingRegion(1, 80000, 1, 90000, 80000, 95000)},
new HashingRegion[] {new HashingRegion(1, 70000, 1, 90000, 65000, 95000)});
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {new HashingRegion(0, 50000, 1, 59999, 49500, 59999)},
new HashingRegion[] {new HashingRegion(0, 50000, 1, 65000, 49500, 65500)});
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {new HashingRegion(1, 80000, 2, 50000, 80000, 50500)},
new HashingRegion[] {new HashingRegion(1, 70000, 2, 50000, 69500, 50500)});
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {
new HashingRegion(0, 50000, 1, 59999, 49500, 59999),
new HashingRegion(1, 80000, 1, 90000, 80000, 90500)
},
new HashingRegion[] {new HashingRegion(0, 50000, 1, 90000, 49500, 90500)});
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {
new HashingRegion(1, 50000, 1, 59999, 49500, 59999),
new HashingRegion(1, 80000, 2, 50000, 80000, 50500)
},
new HashingRegion[] {new HashingRegion(1, 50000, 2, 50000, 49500, 50500)});
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {
new HashingRegion(0, 50000, 1, 59999, 49500, 59999),
new HashingRegion(1, 80000, 2, 50000, 80000, 50500)
},
new HashingRegion[] {new HashingRegion(0, 50000, 2, 50000, 49500, 50500)});
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {new HashingRegion(0, 50000, 1, 59999, 49500, 59999)},
new HashingRegion[] {new HashingRegion(0, 50000, 1, 70000, 49500, 70500)});
checkExclusionAndPadding(
rg,
sequenceNameMap,
new HashingRegion[] {new HashingRegion(1, 80000, 2, 50000, 80000, 50500)},
new HashingRegion[] {new HashingRegion(1, 70000, 2, 50000, 69500, 50500)});
}
