def test_subsample(self):
actual = subsample_paired(self.demux_data, fraction=0.5)
fwd_subsampled_sequence_ids, fwd_obs_sample_count = \
self._validate_fastq_subsampled(actual, self.demux_data,
forward=True)
rev_subsampled_sequence_ids, rev_obs_sample_count = \
self._validate_fastq_subsampled(actual, self.demux_data,
forward=False)
self.assertEqual(fwd_obs_sample_count, 5)
self.assertEqual(rev_obs_sample_count, 5)
# some sequences have been removed - this could occasionally fail,
# but the frequency of that should be ~ 2 * 0.5 ** 11
f_seq_count = self._get_total_sequence_count(
fwd_subsampled_sequence_ids)
r_seq_count = self._get_total_sequence_count(
rev_subsampled_sequence_ids)
self.assertTrue(0 < f_seq_count < 11)
self.assertTrue(0 < r_seq_count < 11)
self.assertEqual(f_seq_count, r_seq_count)
self.assertEqual(fwd_subsampled_sequence_ids,
rev_subsampled_sequence_ids)
def test_correct_output_files_on_small_subsample(self):
# some or all of the output files are likely to be empty, but they
# should still be present and in the manifest
actual = subsample_paired(self.demux_data, fraction=0.00001)
_, fwd_obs_sample_count = \
self._validate_fastq_subsampled(actual, self.demux_data,
forward=True)
_, rev_obs_sample_count = \
self._validate_fastq_subsampled(actual, self.demux_data,
forward=False)
self.assertEqual(fwd_obs_sample_count, 5)
self.assertEqual(rev_obs_sample_count, 5)