def _examine(V1, V2, variants, logprior=None, _calc_lh=None):
E, Es = lh.calc_lh(
*[
common.convert_variant_dict_to_tuple(V)
for V in (variants[V1], variants[V2])
], _calc_lh)
Es -= np.max(Es, axis=1)[:, None]
sep = np.nan * np.ones(len(variants[V1]['var_reads']))[:, None]
persamp = np.hstack((
variants[V1]['var_reads'][:, None],
variants[V1]['total_reads'][:, None],
variants[V1]['vaf'][:, None],
variants[V1]['omega_v'][:, None],
variants[V1]['vaf'][:, None] / variants[V1]['omega_v'][:, None],
sep,
variants[V2]['var_reads'][:, None],
variants[V2]['total_reads'][:, None],
variants[V2]['vaf'][:, None],
variants[V2]['omega_v'][:, None],
variants[V2]['vaf'][:, None] / variants[V2]['omega_v'][:, None],
sep,
Es,
))
post = _calc_posterior(E, _complete_logprior(logprior))
return (persamp, E, post)
def _calc_lh_and_posterior(V1, V2, logprior):
evidence, evidence_per_sample = lh.calc_lh(V1, V2)
posterior = _calc_posterior(evidence, logprior)
return (evidence, posterior)
