Maybe something like a multiplicative aracne network?
mat <- matrix(rnorm(1000), nrow=10)
mi <- knnmi.all(mat)
grn <- aracne.m(mi, 0.15)
There's an additive model as well, though the multiplicative model will presumably identify epistatic regulation(s).
There's alot of options available to you like the previously suggested WGCNA analysis, which I believe is correlative in some sense. WGCNA is common for identifying coexpressed "gene programs". Basically, it calculates the distances betweens genes expression values then groups them according to where they split on a dendrogram. It's a little arbitrary where you cut the dendrogram though, and there's no real guarantee that there is a bonfide regulatory interaction between genes within gene programs. Though it seems likely to some extent that genes with similar function or within close proximity to one another in terms of signaling cascades would be correlated to some degree greater than chance.
modified 3.5 years ago
3.5 years ago by
mforde84 • 1.3k