fitNoDE: Fitting the unrestricted model with EM algorithm
In cz-ye/DECENT: Differential Expression with Capture Efficiency adjustmeNT

fitNoDE

R Documentation

Fitting the unrestricted model with EM algorithm

Fit the DECENT model assuming no differentially-expressed (DE) genes

fitNoDE(data.obs, spikes, spike.conc, use.spikes, CE.range, tau.init,
  tau.global, tau.est, normalize, GQ.approx, maxit, parallel)

`data.obs`	Observed count matrix for endogeneous genes, rows represent genes, columns represent cells.
`spike.conc`	A vector of theoretical count for each spike-in in one cell (Only needed if spikes = `TRUE`).
`use.spikes`	If `TRUE`, use spike-ins to estimate capture efficiencies.
`CE.range`	A two-element vector of the lower limit and upper limit for the estimated range of capture efficiencies (ONLY needed if spikes = `FALSE`, default [0.02, 0.10]).
`tau.init`	initial estimates (intcp,slope) that link Beta-Binomial dispersion parameter to the mean expression.
`tau.global`	whether to use the same tau parameters across cell. Default TRUE
`tau.est`	Methods to estimate tau parameters. The default 'endo' corresponds to using endogeneous genes. Other options are 'none' which means tau.init is not further estimated and 'spikes' corresponds to using spike-ins.
`normalize`	Method for estimating size factors, either 'ML' (maximum likelihood, Ye et al., 2017) or 'TMM' (Robinson et al., 2010).
`GQ.approx`	If `TRUE`, use Gaussian-Quadrature approximation to speed up E-step.
`maxit`	maximum number of iterations for EM algorithm.
`parallel`	If `TRUE`, run DECENT in parallel.
`spike`	Observed count matrix for spike-ins, rows represent spike-ins, columns represent cells. Only needed if spikes = `TRUE`).