Fig. 2
From: Benchmarking computational variant effect predictors by their ability to infer human traits
The performance of 24 computational variant effect predictors in predicting two cholesterol-related phenotypes based on the presence of rare LDLR missense variants. Performance comparisons measured the ability of predictors to infer (A) whether participants were taking the cholesterol-lowering medication atorvastatin (AUBPRC) and (B) circulating LDL-C levels (PCC) based on participant LDLR genotype. Mean performance measures were derived from a 10 k-iteration bootstrap resampling, error bars indicate the 95th percentile confidence interval. In each ranking, predictors that were statistically indistinguishable (FDR ≥ 10%) from the top predictor (AlphaMissense) are indicated