addContrast: fit GLMs to reduced model and test for significance

In MatthiasLienhard/qsea: IP-seq data analysis and vizualization

fit GLMs to reduced model and test for significance

Description

This function fits negative binomial GLMs to reduced models defined either by the "contrast" parameter, or by one or several model coefficients (specified by "coef" parameter) set to zero. Subsequently, a likelihood ratio test is applied, to identify windows significantly dependent on the tested coefficient.

Usage

addContrast(qs,glm,contrast,coef,name,verbose=TRUE, nChunks = NULL, 
    parallel = FALSE )

Arguments

qs	a qseaSet object
glm	a qseaGLM object
contrast	numeric vector specifying a contrast of the model coefficients. This contrast can for example be defined using limma::makeContrasts()
coef	alternatively defines the contrast by coefficient(s) of the model tested to be equal to zero.
name	short descriptive name for the contrast (as "TvsN"), used for examples in columns of result tables
verbose	more messages that document the process
nChunks	fit GLMs in multiple chunks
parallel	use multicore processing

Value

This function returns the qseaGLM object, extended by the fitted coefficients of the reduced GLMs, as well as the test statistics. Note that one qseaGLM object can contain several contrasts.

Author(s)

Mathias Lienhard

See Also

limma::makeContrasts(), fitNBglm(), isSignificant()

Examples

qs=getExampleQseaSet()
design=model.matrix(~group, getSampleTable(qs))
TvN_glm=fitNBglm(qs, design, norm_method="beta")
TvN_glm=addContrast(qs,TvN_glm, coef=2, name="TvN")

MatthiasLienhard/qsea documentation built on March 22, 2023, 1:15 p.m.