CentipedeMixEm-class: Actual EM algorithm for CENTIPEDE

In cmboye/CENTIPEDE_Reimplementation: Reimplementation of CENTIPEDE

Description

Actual EM algorithm for CENTIPEDE

Arguments

k	number of mixed distributions
l	number of rows in X
s	number of columns in X
X	observed experimental data matrix
R	number of reads in each motif, sum of each row in X
G	prior information matrix (PWM+Cons.Score+TSS)
Z	binary variable used to initialize beta's
pi_mat	prior probability matrix
B	beta's for the logistic regression used to calculate pi_l
a0	parameters for negative binomial distribution unbound
a1	parameters for negative binomial distribution bound
tau0	parameters for negative binomial distribution unbound
tau1	parameters for negative binomial distribution bound
lambda	parameters for the multinomial distribution
prob_mat	posterior probability matrix
log_prob_vec	log of posterior probability of bound
max_log	maximum of log posterior bound probability
loglik	log likelihood
tol	threshold for testing convergence
logittau0	logit scale of tau0
logittau1	logit scale of tau1
logitbeta	logit scale of beta

Value

A list of convergence, updated parameters, posterior probabilities, iterations, log likelihood list, bound or unbound status #'@export

cmboye/CENTIPEDE_Reimplementation documentation built on Dec. 23, 2021, 10:19 p.m.