makeModels: Build model matrices for differential expression

View source: R/makeModels.R

makeModelsR Documentation

Build model matrices for differential expression

Description

Builds the model matrices for testing for differential expression by comparing a model with a grouping factor versus one without it. It adjusts for the confounders specified and the median coverage of each sample. The resulting models can be used in calculateStats.

Usage

makeModels(sampleDepths, testvars, adjustvars = NULL, testIntercept = FALSE)

Arguments

sampleDepths

Per sample library size adjustments calculated with sampleDepth.

testvars

A vector or matrix specifying the variables to test. For example, a factor with the group memberships when testing for differences across groups. It's length should match the number of columns used from coverageInfo$coverage.

adjustvars

Optional matrix of adjustment variables (e.g. measured confounders, output from SVA, etc.) to use in fitting linear models to each nucleotide. These variables have to be specified by sample and the number of rows must match the number of columns used. It will also work if it is a vector of the correct length.

testIntercept

If TRUE then testvars is ignored and mod0 will contain the column medians and any adjusting variables specified, but no intercept.

Value

A list with two components.

mod

The alternative model matrix.

mod0

The null model matrix.

Author(s)

Leonardo Collado-Torres

See Also

sampleDepth, calculateStats

Examples

## Collapse the coverage information
collapsedFull <- collapseFullCoverage(list(genomeData$coverage),
    verbose = TRUE
)

## Calculate library size adjustments
sampleDepths <- sampleDepth(collapsedFull,
    probs = c(0.5), nonzero = TRUE,
    verbose = TRUE
)

## Build the models
group <- genomeInfo$pop
adjustvars <- data.frame(genomeInfo$gender)
models <- makeModels(sampleDepths, testvars = group, adjustvars = adjustvars)
names(models)
models

lcolladotor/derfinder documentation built on May 10, 2023, 11:07 p.m.