Skip to main content

Table 3 The most informative patterns for PD138

From: Prediction of DNA-binding proteins from relational features

  Structural Pattern N
1 res(A,arg) 10
2 res(A,arg), res(B,lys), dist(A,B,4.0) 10
3 res(A,arg), res(B,arg), res(C,lys), dist(A,B,10.0), dist(A,C,10.0) 10
4 res(A,arg), res(B,arg), dist(A,B,6.0) 10
5 res(A,arg), res(B,lys), dist(A,B,6.0) 9
6 res(A,ile), res(B,arg), res(C,arg), dist(A,B,6.0), dist(A,C,10.0) 7
7 res(A,leu), res(B,glu), res(C,arg), dist(A,B,10.0), dist(A,C,6.0) 7
8 res(A,lys), res(B,arg), dist(A,B,10.0) 7
9 res(A,arg), res(B,arg), res(C,leu), dist(A,B,10.0), dist(A,C,6.0) 7
10 res(A,arg), res(B,arg), dist(A,B,10.0) 6
  1. The ten most informative structural patterns according to the χ2criterion for the dataset PD138. N is the number of folds, for which the actual pattern was one of the ten best patterns.