Skip to main content

Table 1 NCV experiment results. FATHMM-indel’s performance across 50 data sets created by randomly subsampling the neutral (EVS) class

From: An integrative approach to predicting the functional effects of small indels in non-coding regions of the human genome

Sensitivity (SE) Specificity (SE) Balanced accuracy (SE) AUC (SE)
0.886 (0.005) 0.891 (0.005) 0.889 (0.004) 0.950 (0.003)
  1. The small standard errors (SEs) indicate it is consistent to use one random EVS sample to train the final model