Fig. 3
Two-component pleiotropy score method. We (i) collect association statistics from the UK Biobank, (ii) process them using Mahalanobis whitening, (iii) compute the two components of our pleiotropy score (Pm and Pn) based on the whitened association statistics, (iv) use LD scores to correct for LD-induced pleiotropy (\( {P}_m^{\mathrm{LD}} \) and \( {P}_n^{\mathrm{LD}} \)), and (v) use permutation-based P values to correct for polygenic architecture (\( {P}_m^P \) and \( {P}_n^P \))