Hello!

I am currently using RNA-seq to examine the roles of a few key proteins in the transcriptional response underlying a certain condition. I came across this interesting article (https://www.nature.com/articles/srep12615) where Bayesian Network Analysis (using the R package deal) was used to analyze RNA-seq data to infer which genes are acting "upstream" or "downstream" of their gene of interest. This seems really useful as it would provide me with some potential candidate genes as upstream regulators of a poorly-understood gene in the condition that I am studying, which I can test using knockdowns and etc.

I haven't come across other recent examples of this method of analyzing RNA-seq data. I am wondering if this is a methodologically sound way of identifying potentially causal regulatory relationships between genes that then can be tested using wet lab experiments?

P.S. I tried to look on Biostars for similar threads, but those don't seem to get any answers. I have also emailed the senior author of the paper, and am waiting to hear back.

Thank you very much for any light you can shed on this!

As far as I understand it, bayesian networks are just an alternative to correlation networks (WGCNA) or mutual information networks (ARACNE). I think your false positive rate would probably be high, but it is a valid way of finding candidates, whether it is methodologically sound.... I think that is harder to judge.

Thank you very much for your input!