build_limma: Building a linear regression model for differential analysis

In greenelab/ADAGEpath: ADAGE-based Signature Analysis

Building a linear regression model for differential analysis

Description

Uses limma to build a linear model to test the differential expression or differential activation of either gene features or signature features. Only two-group comparison is implemented for now.

Usage

build_limma(input_data, phenotypes, control_pheno = NULL,
  use.bonferroni = FALSE)

Arguments

input_data	a data.frame that stores either the signature activities or gene expression values. The first column specifies feature names (genes or signatures).
phenotypes	a factor (or a charactor that can be converted into a factor) with two levels that describes the phenotype of each sample.
control_pheno	a character to specify the control phenotype in the limma test. It must be one of the phenotypes level. The other phenotype would be the treatment group. If not specified, the first level of the pehnotypes factor will be used as the control phenotype. (default: NULL)
use.bonferroni	a logical value indicating whether to use the more conservative "bonferroni" method in the p value adjustment. This is recommended when there are too many significant features when using the default "BH" method. (default: FALSE)

Value

a data.frame that stores the result table returned by limma. It includes logFC, adj.P.Val, and other statistics for each feature. LogFC only makes sense when input is gene expression values. If the input is signature activity, then logFC equals to absolute difference. Features with positive logFC have higher values in the second phenotypes level.

See Also

https://bioconductor.org/packages/release/bioc/html/limma.html

greenelab/ADAGEpath documentation built on May 25, 2022, 7:11 a.m.