Diet might still be major: they say their predictive model got ethnicity right 62% of the time, from the bacterial info. I suspect that you could also predict e

Diet might still be major: they say their predictive model got ethnicity right 62% of the time, from the bacterial info. I suspect that you could also predict ethnicity better than chance from diet logs. To really tease out the genetic influence, you'd want to standardize diet, and give all strains equal chance to colonize. Perhaps, compare cohabiting similar-diet same-sex kissing-partners of different-ethnicities, after an initial palate-cleansing course of antibiotics? It's the only way to be sure!