etc/Q-Gen/q-gen-help.md

In lianos/multiGSEA: A unified interface to a plethora of gene set enrichment analysis methods

Generalized qusage of differential gene expression results from a linear mixed model

Description

A wrapper function that implements the qusage algorithm on gene expression results derived from any linear mixed model

Usage

qusage_gen(resids, labels, estimates, dof, std.errors, geneSets, var.equal=TRUE)

Arguments

• resids: An ExpressionSet or matrix containing the corresponding raw residuals obtained from a linear mixed model applied to each row of log2 normalized expression data.

• labels: Vector of labels representing the groupings of the column of resids.

• estimates: Numeric vector of log2 fold change estimates from a single comparison obtained from the linear mixed model. Should correspond to the order in which the genes are present in resids.

• dof: Numeric vector of the degrees of freedom associated with the t-statistic for the corresponding comparison of estimates. Should correspond to the order in which the genes are present in resids.

• std.errors: Numeric vector of the standard errors associated with the t-statistic for the corresponding comparison estimates. Should correspond to the order in which the gene are present in resids.

• geneSets: Either a list of pathways, or a single vector of gene names c orresponding to a single gene set.

• var.equal: Specifying if the linear mixed model assumed equal variances or not. If FALSE, the labels input should correspond to the groupings that were assumed to have unequal variances in the linear mixed model.

Details

This function provides the necessary steps to apply the qusage algorithm to differential gene expression analysis that were conducted using more statistically advanced models. Only the original qusage package is necessary to be installed to run this function.

Note that when random effects are present, it is still up to the user to determine if the residuals matrix provided for the analysis is the conditional residual matrix or a marginal residual matrix treating the random effects as fixed effects.

For large number of random effect replicates, it will not make much difference as the two approaches will converge, however for a small number of replicates, say 2 to 5, we recommend the latter approach in providing the residual matrix.

This decision is solely for the purpose of VIF estimation, the t-statistic information should be given based on the model using fixed and random effects as they were originally specified.

lianos/multiGSEA documentation built on Nov. 17, 2020, 1:26 p.m.