Skip to main content

Table 5 Statistical performance of our model for predicting hotspot of the independent test set by types of amino acids

From: Protein-protein interface hot spots prediction based on a hybrid feature selection strategy

Types of residues Accuracy Specificity Recall Precision F-measure
All residues (28 HS/ 67 NS) 0.705 0.657 0.821 0.500 0.621
Charged residuesa (13 HS/ 28 NS) 0.732 0.750 0.692 0.563 0.621
Non Charged residues 0.685 0.590 0.933 0.467 0.622
Polar residuesb (14 HS/41 NS) 0.709 0.707 0.714 0.455 0.556
Non polar residues 0.700 0.577 0.929 0.542 0.685
Aromatic residuesc (11 HS/12 NS) 0.652 0.417 0.818 0.563 0.667
Non aromatic residues 0.736 0.709 0.824 0.467 0.596
  1. aCharged residues: D,E,K,R,H
  2. bpolar residues: D,E,K,R,H,S,T,N,Q
  3. caromatic residues: F, H,W,Y