Thanks for the comment, I’ll look into this method
Say I have done DESeq2 on my RNA-Seq dataset: experimental vs. control
DESeq2 has a column for BH - adjusted Pvalues and I plan to take the genes with less than < 0.05 adj. pvalue and run it with DAVID and use their functional annotation clustering tool.
I run GSEA (ranked by Signal2Noise and permutated by gene type since each condition had only 6 samples each) with the whole gene list against msigdb C5 GO BP gene sets and had the Enrichment Map app ( with permissive parameters such as p-value <0.05 and FDR q-value <0.25 and a overlap coefficient of 0.5) on Cytoscape to visualize the network. I then used AutoAnnotate to annotate clusters based on frequencies of GO BP words in the clusters.
Can I say that the clusters found in DAVID are significant and compare the GO BP terms found in DAVID to GSEA GO BP terms and do analysis?
1 answer
Since you tagged 'cytoscape', I'll share a common Cytoscape approach to this type of analysis that just uses STRING and related enrichment services:
- Paste your DE genes into the STRING protein search tool in Cytoscape (via stringApp)
- Load your data onto the network
- Update data visualization styles per your taste
- Perform functional enrichemnt analysis using the same stringApp
Now you have enrichment analysis results across multiple up-to-date resources and you can run EnrichmentMap with a click from the stringApp results table as well.
See exercise 1 in this tutorial for more details: https://cytoscape.org/cytoscape-tutorials/protocols/differentially-expressed-genes/#/ex1-up-string
Log in to answer this question.