ztruncnorm: Fit a zero-inflated censored normal distribution.

In JunLiuLab/SIMPLEs: SIMPLEs: single-cell RNA sequencing imputation and cell clustering methods by modeling gene module variation

Description

Fit each gene expression by a zero-inflated censored normal distribution by direct maximize the log-likelihood and return MLE of mean, variance and probability of mixture component.

Usage

ztruncnorm(
  dat,
  cutoff = 0.1,
  iter = 200,
  s_upper = 4,
  s_lower = 0.1,
  p_min = 0.6,
  p_max = 1
)

Arguments

dat	Data matrix.
cutoff	The value below cutoff is treated as no expression. Default = 0.1.
iter	Max number of iteration for optimization. Default = 100.
s_upper	upper bound of standard deviation. Default = 4.
s_lower	lower bound of standard deviation. Default = 0.1.
p_min	lower bound of the probability of the normal component. Default = 0.6.
p_max	upper bound of the probability of the normal component. Can be either a scaler or a vector. If it is a vector, each element is the upper bound for a gene. Default = 1.

Details

It will return NA if number of nonzero is less than 4 indicating that the gene has no expression or if the optimization is not convergence or error which will be filled with sample mean and std, p_max afterwards.

Value

ztruncnorm returns a list of results in the following order.

1. paramA matrix with mean and standard deviation for each gene.

2. pgA vector for the probablity of the normal component.

Author(s)

Zhirui Hu, [e-mail]

Songpeng Zu, [e-mail]

JunLiuLab/SIMPLEs documentation built on March 18, 2021, 3:10 a.m.