Differential Expression Analysis using Limma and qsvaR

Differential Expression Analysis using Limma and qsvaR


Author(s): Hedia Tnani,Joshua M. Stolz,Leonardo Collado-Torres,Louise A. Huuki-Myers,Leonardo Collado Torres

Affiliation(s): Lieber Institute for Brain Development

Social media: https://twitter.com/TnaniHedia

RNA-seq and its expression levels are vulnerable to both environmental and technical influences. In previous studies that explored gene expression data from bulk RNA-seq of postmortem human brain samples, degradation was identified as a crucial and often overlooked issue, particularly when comparing patients to controls. Analyses that do not adjust for degradation effects can lead to false positive differentially expressed genes due to this strong confounder. Furthermore, previous attempts to model RNA quality failed to remove the effects of degradation. To address this problem, the quality surrogate variable analysis (qSVA) method to remove the RNA quality confounding was developed (Jaffe et al., PNAS, 2017). To make it applicable to more brain regions as well as make it user friendly, we developed the qsvaR Bioconductor package. During the workshop, we will provide a step-by-step explanation of the qsvaR Bioconductor package (http://www.bioconductor.org/packages/release/bioc/html/qsvaR.html) for entry-level users and apply it to a publicly available example dataset. We will describe how using qsvaR can improve the reproducibility of DE analyses across datasets. This workshop will also be useful for those interested in learning the basics of limma and differential expression analysis, with a focus on postmortem human brain data.

On YouTube: