This is a test version of Biostars. For the public version, visit https://www.biostars.org.
Comparing GSEA results across multiple cohorts

Hi all,

I performed GSEA on several cohorts using the same pipeline and the same gene set collection (e.g. Hallmark), but separately. Now I would like to identify pathways that are commonly enriched across these cohorts and visualize the results.

However, I’m unsure how to properly compare the outputs, especially the normalized enrichment score (NES). From what I understand, NES is normalized within each collection of dataset, so I’m not sure whether it is valid to compare NES values directly across cohorts.

Ideally I would like to do a heatmap but which value can I use to compare the shared pathways across different cohort ?

gsea heatmap
• 283 views • link
updated by ATpoint ATpoint • written by Picasa Picasa

1 answer

I would not compare GSEA results across cohorts. You have do guarantee that the genes causing the enrichment (so the top genes or "leading edge") are the same between the cohorts. Maybe stats are similar but it is a completely different set of leading edge genes. I would rather do DE per cohort, combine results via meta-analysis (Fisher, Stouffer, Robust Rank Aggregation, anything like that), and run gsea on the meta-rankings.

• 0 views • link
written by ATpoint ATpoint

Log in to answer this question.

More posts like this