Given an observable trait, for example height, how much of it can be explained by genetics as opposed to non-genetic effects? We answer this question by setting up a high-dimensional regression model and estimating the fraction of variance of the trait explained by the high-dimensional predictors, such as SNPs.

It turns out that similar methodology can be used to explain the  fraction of variance of the trait explained by other non-genetic features such as brain images and brain activity maps.