Skip to main content

Advertisement

Table 5 F1 score of individual secondary structure labels using CB513

From: Prediction of 8-state protein secondary structures by a novel deep learning architecture

Q8 Label CRRNNa CRRNN NCCNNa NCCNN MUFOLD-SS DCRNN2 DCRNNa DeepCNF
H 0.903 0.892 0.889 0.884 0.886 0.891 0.880 0.876
B 0.138 0.139 0.089 0.077 0.000 0.006 0.050 0.049
E 0.834 0.814 0.805 0.793 0.789 0.803 0.789 0.788
G 0.463 0.413 0.374 0.360 0.387 0.387 0.317 0.340
I 0 0 0 0 0 0 0 0
T 0.594 0.555 0.565 0.549 0.561 0.550 0.540 0.529
S 0.433 0.397 0.343 0.334 0.373 0.342 0.336 0.335
L 0.660 0.629 0.631 0.621 0.622 0.611 0.610 0.611
macro-F1 0.503 0.480 0.462 0.452 0.452 0.449 0.440 0.441
micro_F 0.74 0.714 0.714 0.704 0.705 0.704 0.697 0.683
  1. aindicates ensemble model
  2. Boldface numbers indicate best performance