Skip to main content

Table 5 Validation of KD4i by 10 fold cross-validation

From: A comprehensive study of small non-frameshift insertions/deletions in proteins and prediction of their phenotypic effects by a machine learning method (KD4i)

Fold Accuracy Sensitivity Specificity Precision NPV MCC
0 77% 87% 67% 76% 81% 0.55
1 83% 93% 74% 78% 91% 0.68
2 78% 85% 71% 76% 81% 0.56
3 79% 96% 63% 72% 94% 0.62
4 78% 88% 68% 74% 85% 0.57
5 73% 86% 60% 67% 82% 0.48
6 77% 85% 71% 73% 83% 0.56
7 82% 90% 74% 79% 87% 0.65
8 83% 85% 81% 83% 83% 0.66
9 79% 90% 65% 76% 85% 0.58
Average 79% 89% 69% 75% 85% 0.59
  1. NPV: negative prediction value, MCC: Matthews Correlation Coefficient.