Skip to main content

Advertisement

Table 4 A comparison of the Q8 accuracy(%) on CB513, CASP10, CASP11 and CASP12 between CRRNN and other state-of-the-art methods

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

method CB513 CASP10 CASP11 CASP12
GSN 66.4 - - -
BLSTM 67.4 - - -
DeepCNF 68.3 71.8 71.7b 0.694b
DCRNN 69.7 - - -
DCRNN2a 70.4 73.9 71.2 68.8
NCCNN 70.3 - - -
NCCNNa 71.4 - - -
MUFOLD-SSb 70.5 74.2 71.6 69.5
CRRNN 71.4 ±0.2 73.8 ±0.5 71.6 ±0.7 68.7 ±0.8
eCRRNNa 74 76.3 73.9 70.7
  1. aindicates ensemble model
  2. bData is generated by our experiment
  3. Boldface numbers indicate best performance