= 27d567bc2d48d9f40e9ccc20ea370b15)
Douglas_Knight comments on Whole genome sequencing vs SNP genotyping - Less Wrong Discussion
You are viewing a comment permalink. View the original post to see all comments and the full post content.
= 783df68a0f980790206b9ea87794c5b6)
Loading…= b715d02e85f7b3b4ae65ca3d3e450868)
Subscribe to RSS Feed
Powered by Reddit
= f037147d6e6c911a85753b9abdedda8d)
You are viewing a comment permalink. View the original post to see all comments and the full post content.
Comments (9)
Whole genome data won't give you health information beyond SNPs because the studies used to generate the health claims use SNPs, because they're cheap.
But there is very little useful data available from SNPs, anyhow. What kind of data do you believe is available and what actions might you take?
If the baseline is a 30% chance of dying of heart disease and it told you that you had a 70% chance, what would you do differently? (but it won't.) Probably you should already be doing it because 30% is a big number!
If the baseline is 0.1% chance of MS and it told you that you have a 0.2% chance of MS, what would you do differently?
The only thing it might tell you that is useful is that you have a high chance of developing what is otherwise a rare disease. It will probably tell you that you have a mildly broken BRCA and have a 22% chance of breast cancer, rather than a 20% chance. That is worthless, even if it is true. But if you're Ashkenazi, there is a 1% chance that it will tell you that you have a fully broken BRCA. That means an 80% chance of breast cancer, which is really not that different from a 20% chance. But it also means aggressive breast cancer at a young age. This is really a different disease. It is a substantial chance of death at a young age. Many women get prophylactic mastectomies. The Dutch have the same 1% base rate as Ashkenazi, but 23andMe does not test for them (at least as of v3). Thus whole genome sequencing can detect the 1% of Dutch with fully broken BRCA and the 1/10,000 of other populations. But those base rates are silly. If you have this gene in your family, you should know already from the high prevalence of breast cancer, especially young fatal breast cancer.
Also, there's a SNP that predicts how fast you metabolize drugs. That sounds useful, but no one takes advantage of it. More generally, it is quite rare that people experiment with doses of drugs, which they really should.
If you do get a whole sequence, you can test every gene to see if it is fully broken ("nonsense mutation" as opposed to the mild "missense"). If you have such a gene, it probably is a big deal, but it will probably be hard to interpret.
That line of argument is flawed because actions have costs.
To give a simple example (and discussing whether it's precisely correct is besides the point), you can take baby aspirin to reduce the risk of heart attacks. Aspirin is a blood thinner, it makes clots (which cause heart attacks) less likely. However decreasing coagulation is not an unmitigated blessing. If you get internal bleeding -- e.g. a blood vessel ruptured in your brain -- that aspirin you've been taking could make things significantly worse. It's a trade-off.
Given this, you want to know on which side (basically, heart attacks vs. strokes) is your personal risk the highest. It is actionable knowledge.
Looks very different to me -- you are quite cavalier with a fourfold difference in odds...
4 is a small number. It is pretty rare that a cost-benefit calculation cares about that factor of 4, that multiplying the benefits by 4 will change the decision from reject to accept.
It seems we have a different perception of smallness and tend to encounter different cost-benefit calculations.