RA3_EM: EM algorithm for RA3

In cuhklinlab/RA3: The R-based implementation of "A reference-guided approach for epigenetic characterization of single cells".

Description

This function is implementing an EM algorithm to estimate parameters of RA3 model.

Usage

RA3_EM(Y, K1, K2, K3, Gamma, A, W, sigma_s)

Arguments

Y	Input single cell count matrix.
K1, K2, K3	Number of components consisted in RA3 model.
Gamma	Initial matrix of parameter matrix Γ.
A	Initial matrix of precision matrix A.
W	Initial matrix of parameter matrix W, a warm start is recommended in runRA3.
sigma_s	Intial value of parameter σ^2, a warm start is recommended in runRA3.

Value

A list containing the following components:

H	the extracted latent features H.
W	the estimated parameter matrix W.
Beta	the estimated covariance parameter vector β.
Gamma	the estimated indicator matrix Γ.
A	the estimated precision matrix A.
sigma_s	the estimated σ^2.
lgp	the largest log posterior value when EM algorithm converges.

cuhklinlab/RA3 documentation built on March 18, 2021, 4:38 p.m.