Figure 5
From: A machine-learning heuristic to improve gene score prediction of polygenic traits
GraBLD polygenic risk score discrimination as a function of calibration set sample size. The size of the UKB calibration set varied from 20 to 10,000 for height and BMI, and from 3 to 1,000 case-control pairs for diabetes. For each calibration sample size, discrimination of the corresponding polygenic risk score was calculated in the independent UKB validation set (N = 130,215 for height and BMI; N = 5,746 case-control pairs for diabetes).
