Skip to main content

Table 4 The performance of our model and the model using the contact number

From: Prediction of conformational B-cell epitopes from 3D structures by random forests with a distance-based feature

  Bound dataset Unbound dataset
Model AUC F ACC SN SP AUC F ACC SN SP
Contact Number 0.565 0.188 0.621 0.468 0.638 0.619 0.241 0.573 0.601 0.558
ARD 0.589 0.206 0.620 0.487 0.633 0.627 0.243 0.585 0.578 0.588
CF+CN 0.618 0.212 0.670 0.444 0.695 0.649 0.259 0.631 0.572 0.634
CF+ARD 0.633 0.227 0.672 0.490 0.692 0.654 0.256 0.608 0.601 0.606
CF+ARD+CN 0.623 0.215 0.654 0.481 0.6717 0.646 0.243 0.593 0.587 0.589
  1. Contact Number and ARD mean the models using the contact number and ARD. CF+CN represents the model based on the surface patch by combining the conventional features and contact number; CF+ARD represents our model based on the thick surface patch by combining the conventional features and ARD. CF+ARD+CN means the altered edition of our method that incorporates the contact number.