>>12651104Not that person but it's a population structure problem.
>Gene varients A and B are found to relate strongly to cancer in GWAS >Publish bunch of papers on it>Discover A and B literally do nothing>Look at data and notice that the A and B clusters were correlated on a lot of genes>Look up where those people lived>Discover it's because there was a bunch of loosely related families living next to a chemical factory. But instead you have relationships between facial structure which are far less easily controlled or quantified. So you end up just going:
>Pronounced brow ridge seen in those who drink excessively in my sample>The pronounced brow ridge is over represented in native american groups >Cultural aspects combined with other underlying gene patterns, unrelated to facial structure, mean that they are more vulnerable to alcoholism>Wrongly assume that some random person you meet is an alcoholic because of their brow ridge 
