Robert M. Porsch

This paper develops methods to calculate polygenic scores using raw genetic data from UK Biobank while avoiding overfitting, showing improved predictive power for six phenotypes.

KS-Burden assesses distributional differences of rare variants between affected and unaffected subjects, outperforming SKAT-O in scenarios with causal variant clusters.

This paper introduces lassosum, a penalized regression method for constructing polygenic scores using summary statistics and a reference panel, showing improved accuracy and efficiency compared to existing methods.