Interpreting a list of Differentially Expessed Genes

Question

[ajp104]

Any help anyone can offer is greatly appeciated, I’m in at the deep end with a university project on Bioinformatics that’s completely unrelated to my prior study, and I’m struggling.

So I have a list of ~1000 DEGs (FDR-corrected p value <0.05) from a large scale study into Heart Failure, comparing survivor and non-survivor groups. I’m supposed to be analysing this for pathways and biological mechanisms underpinning mortality.

I’ve got as far as parachuting the list into DAVID and Enrichr, both of which flag up the same three or four pathways on KEGG, and then I’m stuck. How do I go about understanding the role of these pathways and the intrinsic effects of my genes of interest within them? What other ways can I interpret and analyse this kind of list?

Any help or advice is greatly appreciated, I’ve been trying to self-teach a lot of this stuff but it’s slow going and deadlines are approaching.

[Buffo]

What do you expect to have after gsea analysis? I mean, three or four pathways look pretty good summarized result taking into account you have 1000 DE-genes.

Answer 1

[jared.andrews07]

I mean, this is where the bio in "bioinformatics" comes into play. Do the pathways make sense? Read up on the genes involved, how might their deregulation impact heart failure? Do they tie together in a way that could be tested experimentally, either in cell lines or animal models or other assays? What could the changes mean mechanistically?

Are the genes in these pathways all up-regulated? Down-regulated? What enriched pathways/ontologies do you get if you do these analyses for your up-regulated and down-regulated gene lists separately?

This is where knowledge regarding your disease of interest and a certain amount of manual curation are key in explaining the differences you're observing. But it's also one of the more enjoyable parts of bioinformatics, in my opinion, as you get to tie your results back to something biologically tangible. Develop some hypotheses based on your results, figure out how to test them, and then do so, if possible. There is no easy answer.