Based on this research we can say that the gene isn't necessarily responsible, if at all even relevant. It does indicate association, and that's where it ends. They made a probability correlation, layered on top of a subjective pool of symptoms. For those that have never been involved in any research, ask yourself, how could you NOT find a correlation when your subject set is in the billions, all sharing millions of genes in common? You'll find a gene correlated with car accidents if you make assumptions about data like this headline suggests. There is very little interesting here. This is just scratching the surface of a much larger study that is highly likely to also turn up nothing of interest.
Welcome to genomics. But based on your handle, I'm guessing you already work in a genomics-related area.
Genomics has a terrible hype problem, like ML. Genomics researchers have decades of experience making their papers souond far more significant and actionable than they really are.
We've mostly moved past the "we found a gene for..." which sort of considers phenotypes as mendelian (only two alleles per gene, independent segregation, single SNP is highly penetrant, etc) to the association approach, which as you point out, isn't satisfying from a causality or mechanistic viewpoint.
In adopting the association approach we have learned (unsurprisingly IMHO) that most organismal phenotypes (I'm treating this risk as a phenotype, which is a fairly loose interpretation) are caused by the interaction of the environment and thousands of changes to thousands of genes, all of which are non-additive and non-linear.
Yes. Agreed. I think the headline is trying to make it sound like something like this was discovered, but that's far from the level of progress described in the actual research. It's the excess advertising that's ruining peoples' view of science.
