def summarize_variations(in_zarr_path,
out_dir_path,
draw_missin_rate=True,
draw_mac=True,
draw_maf=True,
draw_obs_het=True,
min_call_dp_for_het_call=MIN_DP_FOR_CALL_HET,
min_num_genotypes=MIN_NUM_GENOTYPES_FOR_POP_STAT,
num_bins=DEF_NUM_BINS,
silence_runtime_warnings=True):
stats = {}
variations = load_zarr(in_zarr_path)
max_alleles = variations[ALT_FIELD].shape[1]
num_variations = variations.num_variations
num_samples = variations.num_samples
if draw_missin_rate:
_stats = calc_called_gt(variations, rates=True)
counts, edges = va.histogram(_stats, n_bins=num_bins, limits=(0, 1))
stats['called'] = {'counts': counts, 'edges': edges}
if draw_mac:
_stats = calc_mac(variations, max_alleles, min_num_genotypes)
counts, edges = va.histogram(_stats,
n_bins=num_bins,
limits=(0, variations.num_samples))
stats['mac'] = {'counts': counts, 'edges': edges}
if draw_maf:
_stats = calc_maf_by_gt(variations, max_alleles, min_num_genotypes)
counts, edges = va.histogram(_stats, n_bins=num_bins, limits=(0, 1))
stats['maf'] = {'counts': counts, 'edges': edges}
if draw_obs_het:
_stats = calc_obs_het(
variations,
min_num_genotypes=min_num_genotypes,
min_call_dp_for_het_call=min_call_dp_for_het_call)
counts, edges = va.histogram(_stats, n_bins=num_bins, limits=(0, 1))
stats['obs_heterocigosity'] = {'counts': counts, 'edges': edges}
computed_stats = compute(stats,
silence_runtime_warnings=silence_runtime_warnings)
for kind, stats in computed_stats.items():
with (out_dir_path / f'{kind}.png').open('wb') as out_fhand:
plot_histogram(stats['counts'],
stats['edges'],
out_fhand,
log_scale=True)
with (out_dir_path / 'stats.txt').open('w') as fhand:
fhand.write(f'STATS FOR: {in_zarr_path.name}\n')
fhand.write('-----------' + '-' * len(in_zarr_path.name) + '\n')
fhand.write(f'Num. variations: {num_variations}\n')
fhand.write(f'Num. samples: {num_samples}\n')
fhand.write('\n')
def remove_low_call_rate_samples(variations,
min_call_rate,
rates=True,
filter_id='sample_call_rate',
calc_histogram=False,
n_bins=DEF_NUM_BINS,
limits=None):
num_missing_gts = calc_missing_gt_per_sample(variations, rates=rates)
if rates:
num_called = 1 - num_missing_gts
else:
num_called = utils_array.get_shape_item(variations.gt,
0) - num_missing_gts
selected_samples = num_called >= min_call_rate
variations = keep_samples_with_mask(variations, selected_samples)[FLT_VARS]
num_selected_samples = va.count_nonzero(selected_samples)
num_filtered_samples = va.count_nonzero(va.logical_not(selected_samples))
flt_stats = {
N_SAMPLES_KEPT: num_selected_samples,
N_SAMPLES_FILTERED_OUT: num_filtered_samples
}
if calc_histogram:
limits = (0, 1) if rates else (0, len(variations.num_variations))
counts, bin_edges = va.histogram(num_called,
n_bins=n_bins,
limits=limits)
flt_stats[COUNT] = counts
flt_stats[BIN_EDGES] = bin_edges
flt_stats[HIST_RANGE] = [min_call_rate]
return {FLT_VARS: variations, FLT_ID: filter_id, FLT_STATS: flt_stats}
def filter_by_mac(variations,
max_alleles,
max_allowable_mac=None,
min_allowable_mac=None,
filter_id='filter_by_mac',
min_num_genotypes=MIN_NUM_GENOTYPES_FOR_POP_STAT,
calc_histogram=False,
n_bins=DEF_NUM_BINS,
limits=None):
macs = calc_mac(variations,
max_alleles=max_alleles,
min_num_genotypes=min_num_genotypes)
# print(compute(macs))
result = _select_vars(variations, macs, min_allowable_mac,
max_allowable_mac)
if calc_histogram:
if limits is None:
limits = (0, variations.num_samples)
counts, bin_edges = va.histogram(macs, n_bins=n_bins, limits=limits)
result[FLT_STATS][COUNT] = counts
result[FLT_STATS][BIN_EDGES] = bin_edges
limits = []
if min_allowable_mac is not None:
limits.append(min_allowable_mac)
if max_allowable_mac is not None:
limits.append(max_allowable_mac)
result[FLT_STATS]['limits'] = limits
return {
FLT_VARS: result[FLT_VARS],
FLT_ID: filter_id,
FLT_STATS: result[FLT_STATS]
}
def remove_low_call_rate_vars(variations,
min_call_rate,
rates=True,
filter_id='call_rate',
calc_histogram=False,
n_bins=DEF_NUM_BINS,
limits=None):
num_missing_gts = calc_missing_gt(variations, rates=rates)
if rates:
num_called = 1 - num_missing_gts
else:
num_called = variations.gt.shape[1] - num_missing_gts
selected_vars = num_called >= min_call_rate
variations = variations.get_vars(selected_vars)
num_selected_vars = va.count_nonzero(selected_vars)
num_filtered = va.count_nonzero(va.logical_not(selected_vars))
flt_stats = {N_KEPT: num_selected_vars, N_FILTERED_OUT: num_filtered}
if calc_histogram:
limits = (0, 1) if rates else (0, len(variations.num_samples))
counts, bin_edges = va.histogram(num_called,
n_bins=n_bins,
limits=limits)
flt_stats[COUNT] = counts
flt_stats[BIN_EDGES] = bin_edges
flt_stats['limits'] = [min_call_rate]
return {FLT_VARS: variations, FLT_ID: filter_id, FLT_STATS: flt_stats}
def test_calc_maf_by_gt2(self):
variations = load_zarr(TEST_DATA_DIR / 'test.zarr')
mafs = calc_maf_by_gt(variations, max_alleles=3, min_num_genotypes=0)
# res = compute(mafs, silence_runtime_warnings=True)
counts, edges = va.histogram(mafs, n_bins=5, limits=(0, 1))
cc = compute({
'counts': counts,
'edges': edges
},
silence_runtime_warnings=True)
self.assertTrue(np.all(cc['counts'] == [0, 0, 4, 2, 0]))
self.assertTrue(
np.all(np.isclose(cc['edges'], [0, 0.2, 0.4, 0.6, 0.8, 1])))
def filter_by_obs_heterocigosis(
variations,
max_allowable_het=None,
min_allowable_het=None,
min_call_dp_for_het_call=None,
max_call_dp_for_het_call=None,
filter_id='obs_het',
min_num_genotypes=MIN_NUM_GENOTYPES_FOR_POP_STAT,
calc_histogram=False,
n_bins=DEF_NUM_BINS,
limits=None):
obs_het = calc_obs_het(variations,
min_num_genotypes=min_num_genotypes,
min_call_dp_for_het_call=min_call_dp_for_het_call,
max_call_dp_for_het_call=max_call_dp_for_het_call)
result = _select_vars(variations,
obs_het,
min_allowable=min_allowable_het,
max_allowable=max_allowable_het)
if calc_histogram:
if limits is None:
limits = (0, 1)
counts, bin_edges = va.histogram(obs_het, n_bins=n_bins, limits=limits)
result[FLT_STATS][COUNT] = counts
result[FLT_STATS][BIN_EDGES] = bin_edges
limits = []
if min_allowable_het is not None:
limits.append(min_allowable_het)
if max_allowable_het is not None:
limits.append(max_allowable_het)
result[FLT_STATS]['limits'] = limits
return {
FLT_VARS: result[FLT_VARS],
FLT_ID: filter_id,
FLT_STATS: result[FLT_STATS]
}