likelihood_criteria: Compute pvalues and aic difference criteria

In strayMat/lineageDE: Differentially expressed genes in cell lineages

Description

Perform a regression with reg_vgam or reg_loess function and compute p-values based on a likelihood ratio test as well as a AIC based criteria (computation of the difference of AIC between the null model and the alternative model).

Usage

likelihood_criteria(data, gene, reg.f = "loess", pval = F, span = 0.75,
  s.df = 4, fam = "gaussian", dtw = F)

Arguments

data	a warpDEDataSet with results to be plotted.
gene	character, a gene of interest.
reg.f	the regression method to use, either "loess" with the gam package, "ns" which fits a natural cubic spline with gam package or "splines" which fits a smoothing spline (default is "loess").
pval	logical, if the likelihood criteria is to be computed or not (default is FALSE).
span	numeric, a smoothing parameter for the regression function (default is 0.75, see gam::lo for details).
s.df	numeric, a smoothing parameter for the nsplines regregression (default is 4, see splines::s for details about regularization).
fam	character, for the vgam regression, the distribution assumption of the residuals; etiher "binomial" or "gaussian" (default is "gaussian").
dtw	logical, if the alignment should be made with dtw before acomputing the likelihood, (default is FALSE) .

Value

returns the pvalue and if asked the aic difference for the gene of intereset.

strayMat/lineageDE documentation built on May 30, 2019, 8:18 a.m.