new_majiq.DPsiPrior

class rna_majiq.DPsiPrior(a=[1.0, 75.0, 1000.0], pmix=[0.2, 0.5, 0.3])

Prior on DeltaPsi as weighted mixture of beta distributions (over [-1, 1])

Parameters:

a, pmix (Union[List[float], xr.DataArray]) – Default parameters for prior. Must have same length. If xr.DataArray, must have dimension mixture_component. a is the parameters for each component beta. pmix is the probability of each component.

__init__(a=[1.0, 75.0, 1000.0], pmix=[0.2, 0.5, 0.3])

Methods

__init__([a, pmix])
discretized_logpmf([psibins, PSEUDO])	Get discretized logprior for deltapsi with 2 * psibins bins
empirical_update(psi1, psi2[, minreads, ...])	Use reliable binary events from psi1,2 to return updated prior
empirical_update_EM(dpsi, a, pmix, ...)
fit_a(dpsi, pmix_given_dpsi[, force_slab, ...])	Fit a using dpsi, pmix_given_dpsi (M-step) by method of moments
fit_pmix(pmix_given_dpsi[, pmix_eps])	Fit pmix using pmix_given_dpsi (M-step)
get_empirical_dpsi(psi1, psi2[, minreads, ...])	Get high confidence empirical deltapsi from input groups of experiments
infer_pmix_given_dpsi(dpsi, a, pmix)	Get probability of membership to mixture given observaion (E-step)
legacy_empirical_replace(dpsi[, pmix_mask])	This isn't an update so much as replacement