R-squared statistic measured in the data used to train the model (), estimated posterior mean of heritability (), and Deviance Information Criterion (DIC) by model and number of SNPs (where K = 1,000).

Estimates were obtained by fitting models to height adjusted by sex and age and using all available data (N = 5,117).1For the Bayesian LASSO, due to high memory requirements, only models including up to 80K markers were considered. This model does not include a genetic variance parameter, therefore it does not yield a direct estimate of heritability. For this reason heritability is not reported for this model.