Genomic Prediction of Complex Traits

Steve Hsu discusses using machine learning and biobank data to construct genomic predictors for complex human traits like height and disease risk.

Presented at Michigan State University on March 12, 2018 Steve Hsu Vice President for Research and Graduate Studies vprgs.msu.edu Department of Physics and Astronomy pa.msu.edu/ Abstract: After a brief review (suitable for physicists and engineers) of computational genomics and complex traits, I describe recent progress in this area. Using methods from Compressed Sensing (L1-penalized regression; Donoho-Tanner phase transition with noise) and the UK BioBank dataset of 500k SNP genotypes, we construct genomic predictors for several complex traits. Our height predictor captures nearly all of the predicted SNP heritability for this trait -- thereby resolving the missing heritability problem. Actual heights of most individuals in validation tests are within a few cm of predicted heights. I also discuss application of these methods to cognitive ability and polygenic disease risk: sparsity estimates (of the number of causal loci), combined with phase transition scaling analysis, allow estimates of the amount of data required to construct good predictors. Finally, I discuss how these advances will affect human health and reproduction (embryo selection for In Vitro Fertilization, genetic editing) in the coming decade.