One of the most exciting and thought-provoking papers that I’ve read in the field of human genetics in the past five years is the Omnigenic model. There are several commentaries on the model already, for example here and here.

Once during a class, I asked a professor studying the genetics of complex traits what would be a good stopping point for his GWAS(s) or when would he consider his research question answered. I don’t think he had a good answer. The trend seemed to be collect as many samples as you can, do the GWAS, then argue that the number of samples weren’t enough to find all the loci and do more GWAS. But when will we have all the loci? The top significant hits explain a tiny fraction of the heritability. That was a bit worrying as a grad student coming in to study human and statistical genetics. It is an unwritten rule that if a GWAS finds one significant hit that uncover a new gene that could be a potential drug target the GWAS would be considered a success. We could then forget about the other significant hits or let the ‘experimentalists’ worry about what they are. I don’t know if there’s anyone following up on these other hits, almost all in noncoding regions, perhaps in the secret labs of drug companies but certainly not in the research labs that I get to hear about.

One of my favorite blog-posts makes an analogy comparing the GWAS era to the era of spotting stars in the sky without a theory of gravitation. What are all these GWAS hits? How do they function? We don’t have evidence that all of these are eQTLs, in-fact it looks like only a small fraction of them are eQTLs. What on earth are these loci? My advisor wants to know how do we study such small effects in the lab/should we even bother studying them? How do all these small effects add up to cause the trait? Is there a piece(or many pieces) of gene regulation that we don’t know about that could provide answers?

Some of the criticism against the omnigenic model is that omnigenicity is no different from polygenicity or that the idea of core genes is vague. I agree with the authors that the new term omnigenic helps with a newer conceptual understanding. There are good examples of core genes in GWAS hits that we’ve found, for example in AMD. The omnigenic model doesn’t necessarily provide all the answers, it actually makes us ask questions of the GWAS results to date. I love the fact that Jonathan Pritchard and colleagues are not just adding more loci to the GWAS pot but are taking a step back and trying to explain the findings that we have so far. It just makes sense to do this. Are gene networks the answer? How can we study gene networks with our expanded genome engineering toolkit? Questions abound as the genetics revolution marches on. As a computational geneticist I sometimes just can’t stop thinking about this problem. I’d love to chat over coffee or beer if anyone’s interested in talking more about this in person.