def burden_test(count, pred, offset, test_method, model, s):
""" Perform burden test.
Args:
count:
pred:
offset:
test_method:
model:
s:
use_gmean:
Returns:
"""
if test_method == 'auto':
test_method = 'binomial' if model['pval_dispersion'] > 0.05 else 'negative_binomial'
if test_method == 'negative_binomial':
logger.info('Using negative binomial test with s={}, theta={}'.format(s, model['theta']))
theta = s * model['theta']
pvals = np.array([negbinom_test(x, mu, theta, o)
for x, mu, o in zip(count, pred, offset)])
elif test_method == 'binomial':
logger.info('Using binomial test')
pvals = np.array([binom_test(x, n, p, 'greater')
for x, n, p in zip(count, offset,
pred/offset)])
else:
logger.error('Unknown test method: {}. Please use binomial, negative_binomial or auto'.format(test_method))
sys.exit(1)
return pvals
python类binom_test()的实例源码
def pre(self, df):
if self.variable not in df:
df.loc[:, self.variable] = df.apply(lambda row: binom_test(np.round(row['s1_area_fraction_top'] * row['s1']),
np.round(row['s1']),
self.aft_map(np.sqrt(row['x']**2 + row['y']**2),
row['z'])[0, 0]),
axis=1)
return df
def binomial_test(mut_on_kmer, mut_on_gene, e):
'''
performs binomial tests for mutations on each kmer, returns pvalue
'''
pval = binom_test(mut_on_kmer, mut_on_gene, e)
return(pval)
