single pathway score

Question

[joha]

Hello, I would like to find out a method for "single pathway" differential analysis (if it exists). For example, our team is focused in KEGG pathway hsa03018 (RNA degradation). Our transcriptome data (Treatment vs Control) underwent differential analysis as usual, and GSEA: I happily obtained a statistically significant NES for my pathway of interest. However, I was wondering : Is there any method (even if not yet a code, but at least as scientific paper) to start from the count matrix (Treatment, Control), take the submatrix corresponding to the genes belonging to the pathway hsa03018 ( ~79 genes) and compare the "Treatment set expression" against the "Control set expression"? I have big interest because we have lots of transcriptomes with many different conditions and we are only interested in a maximum of 4-5 pathways. Thank you for any help.

Answer 1

[ATpoint]

Sounds pretty much exactly what https://bioconductor.org/packages/release/bioc/html/GSVA.html does, so transforming a per-gene count matrix into a per-pathway count matrix so that you can use approaches like limma on pathway rather than gene level.