I love that you brought up the case of HDL, such a huge fan of George Davey Smith's work on this topic (https://pubmed.ncbi.nlm.nih.gov/37814277/ & https://pubmed.ncbi.nlm.nih.gov/37814277/). I basically think a lot of new perturbational technologies across many areas of biology and neuroscience could result in the creation/discovery of new instrumental variables and do for other areas what mendelian randomization has done for genetics. But they have to take the imperfections, off-target effects of the perturbation seriously!
Sometimes purported genetic instruments are often not good or not appropriately analyzed. I've been digging into the vitamin-D supplementation fiasco recently. I know very few scientists who both see the long-term promise of a tool and rigorous about acknowledging when they make a mistake. https://www.thelancet.com/journals/landia/article/PIIS2213-8587(23)00348-0/fulltext
I love that you brought up the case of HDL, such a huge fan of George Davey Smith's work on this topic (https://pubmed.ncbi.nlm.nih.gov/37814277/ & https://pubmed.ncbi.nlm.nih.gov/37814277/). I basically think a lot of new perturbational technologies across many areas of biology and neuroscience could result in the creation/discovery of new instrumental variables and do for other areas what mendelian randomization has done for genetics. But they have to take the imperfections, off-target effects of the perturbation seriously!
Sometimes purported genetic instruments are often not good or not appropriately analyzed. I've been digging into the vitamin-D supplementation fiasco recently. I know very few scientists who both see the long-term promise of a tool and rigorous about acknowledging when they make a mistake. https://www.thelancet.com/journals/landia/article/PIIS2213-8587(23)00348-0/fulltext
> And that’s how we can prove causation from correlation in genetics, subject to a key assumption that we know all possible confounders.
This is not a proof, except in a scientism sense of the word.