Skip to main content

Table 4 Classification performance measurements from three tools (%).

From: In silico comparative characterization of pharmacogenomic missense variants

Classification Sensitivity Specificity Accuracy Precision F-Measure AUC
SIFT       
Disease/neutral 71.1 71.1 71.1 75.1 73.1 77.3
   PGx/disease 71.1 60.6 65.8 90.3 79.6 70.5
   PGx/neutral       54.8
PolyPhen2       
Disease/neutral 85.3 57.1 71.2 72.7 78.5 76.9
   PGx/disease 85.3 54.0 69.6 90.6 87.8 69.9
   PGx/neutral       53.8
MutPred       
Disease/neutral 98.5 76.6 87.5 84.9 91.2 95.1
   PGx/disease 98.5 61.1 79.8 92.9 95.6 85.9
   PGx/neutral       59.1
  1. Classification thresholds between PGx and neutral variants are unknown and thus only AUC was computed for each tool.