glmQLFTest

glmQLFTest is a function in the edgeR package (Bioconductor) that performs a quasi-likelihood F-test for differential expression within a generalized linear model framework using RNA-seq count data. It is part of the quasi-likelihood (QL) pipeline designed to provide robust inference when estimating dispersion and testing contrasts among experimental groups.

The test operates on a fitted quasi-likelihood GLM. After preparing a DGEList object with counts, performing normalization, and estimating dispersion, the user fits a model with glmQLFit and then applies glmQLFTest to test a specified coefficient or a linear contrast of coefficients in the design matrix. The test can be specified by indicating a single coefficient (coef) or by providing a contrast vector that encodes the comparison of interest. The result reflects differences in expression associated with the tested effect, accounting for the quasi-likelihood dispersion estimate.

The output of glmQLFTest is an object suitable for downstream summarization, typically viewed with topTags. Each row corresponds to a gene and includes the estimated log2 fold change, the quasi-likelihood F-statistic or related test statistic, the P-value, and the FDR-adjusted P-value. The quasi-likelihood framework is designed to maintain appropriate error rates, particularly in small sample sizes, and is often preferred over likelihood-ratio tests in edgeR for differential expression analyses.

In practice, glmQLFTest provides a targeted, robust approach to testing differential expression across predefined contrasts, fitting into edgeR’s GLM-based workflow for count data analysis.