evalLogLikPiZINB_ManyCells: Cost function zero-inflated negative binomial model for...

In YosefLab/LineagePulse: Differential expression analysis and model fitting for single-cell RNA-seq data

Description

Log likelihood of zero inflated negative binomial model. This function is designed to allow numerical optimisation of logistic drop-out paramater model on multiple cells given the negative binomial mean and dispersion parameters. This function is optimised for memory usage vs evalLogLikPiZINB_SingleCell at the cost of computation time: The parameter models are not kept as a gene x cell matrix but as the raw model objects which are evaluated within the cost function.

Usage

evalLogLikPiZINB_ManyCells(vecTheta, matCounts, lsMuModel, lsDispModel,
  lsDropModel)

Arguments

vecTheta	(numeric vector length linear model) Parameter estimates for logit linear model for drop-out rate.
matCounts	(count matrix genes x cells) Observed read counts, not observed are NA.
lsMuModel	(list) Object containing description of gene-wise mean parameter models.
lsDispModel	(list) Object containing description of gene-wise dispersion parameter models.
lsDropModel	(list) Object containing description of cell-wise drop-out parameter models.

Value

scaLogLik (scalar) Value of cost function: zero-inflated negative binomial likelihood.

Author(s)

David Sebastian Fischer

See Also

Called by fitting wrapper: fitPi_ManyCells. Calls evalLogLikMatrix. Compiled version: evalLogLikPiZINB_ManyCells_comp.

YosefLab/LineagePulse documentation built on May 6, 2019, 2:19 p.m.