def test_downsample_to_too_large_target_count(self):
""" Should fail """
temp_dir = tempfile.mkdtemp()
target_count = 20000
with self.assertRaises(ValueError):
read_utils.main_downsample_bams([self.larger_bam, self.smaller_bam], temp_dir, specified_read_count=target_count, JVMmemory="1g")
def test_downsample_to_target_count_without_subdir(self):
target_count = 4000
read_utils.main_downsample_bams([self.larger_bam], out_path=None, specified_read_count=target_count, JVMmemory="1g")
output_bams = list(glob.glob(os.path.join(os.path.dirname(self.larger_bam), '*downsampled-*.bam')))
print(output_bams)
self.assertGreater(len(output_bams), 0, msg="No output files matching *downsampled-*.bam found")
for out_bam in output_bams:
self.assertAlmostEqual(self.samtools.count(out_bam), target_count, delta=10, msg="{} not downsampled to the target size: {}".format(os.path.basename(out_bam),target_count))
def test_downsample_with_dedup_before(self):
""" Also tests subdir output """
temp_dir = tempfile.mkdtemp()
target_count = 1500
read_utils.main_downsample_bams([self.with_dups], temp_dir, deduplicate_before=True, specified_read_count=target_count, JVMmemory="1g")
output_bams = list(glob.glob(os.path.join(temp_dir, '*.bam')))
self.assertGreater(len(output_bams), 0, msg="No output found")
for out_bam in output_bams:
self.assertAlmostEqual(self.samtools.count(out_bam), target_count, delta=10, msg="{} not downsampled to the target size: {}".format(os.path.basename(out_bam),target_count))
def test_normalization_to_lowest_cardinality(self):
""" Also tests subdir output """
temp_dir = tempfile.mkdtemp()
target_count = self.samtools.count(self.smaller_bam)
# target count not passed in since we are checking that the count of the smaller file is used
read_utils.main_downsample_bams([self.larger_bam, self.smaller_bam], temp_dir, JVMmemory="1g")
output_bams = list(glob.glob(os.path.join(temp_dir, '*.bam')))
self.assertGreater(len(output_bams), 0, msg="No output found")
for out_bam in output_bams:
self.assertAlmostEqual(self.samtools.count(out_bam), target_count, delta=10, msg="{} not downsampled to the target size: {}".format(os.path.basename(out_bam),target_count))