Introduction

Here we compare the performances of different genetic risk prediction (GRIP) methods via a set of simulation studies.

10-fold cross validation with the criteria area under the curve (AUC) is used to assess the performance.

We simulated 5000 individuals with dichotomous traits.

• The liability threshold model is used.
• We set the prevalence to 0.1.

27059 SNPs are included.

• Simulated on chromosome 14.
• 1508 markers are excluded based on HWE test (p ≤ 0.05).
• 4060 markers are excluded based on frequency test (MAF < 0.1).