Skip to main content
. 2020 May 23;9(5):giaa052. doi: 10.1093/gigascience/giaa052

Table 5:

hyperSMURF and parSMURF (with grid search and Bayesian optimization) prediction performance obtained over a fully independent Mendelian test set composed of newly annotated pathogenic variants (positive examples) and common neutral variants (negative examples)

Model AUROC AUPRC
hyperSMURF 0.945216 0.098153
parSMURF, grid search 0.941026 0.409067
parSMURF, Bayesian optimizaion 0.928158 0.192568