Skip to main content

Table 2 Performance comparison between the proposed and the competing methods based on the 7 fold cross validation test on the BT426 dataset

From: Prediction of beta-turns at over 80% accuracy based on an ensemble of predicted secondary structures and multiple alignments

Prediction method [reference] Qtotal Qpredicted Qobserved MCC
This paper 80.9 62.7 55.6 0.47
SVM [33] 79.8 55.6 68.9 0.47
MOLEBRNN [28] 77.9 53.9 66.0 0.45
SVM (multiple alignment) [32] 77.3 53.1 67.0 0.45
BTSVM [31] 78.7 56.0 62.0 0.45
BETATPRED2 (multiple alignment) [26, 27] 75.5 49.8 72.3 0.43
COUDES (ψthreshold = 0 for PSSM) [4] 74.8 48.8 69.9 0.42
COUDES (ψthreshold = -100 for PSSM) [4] 75.5 49.8 66.6 0.41
SVM (single sequence) [32] 74.8 49.1 67.9 0.41
BETATPRED2 (single sequence) [26, 27] 74.3 48.4 71.2 0.41
KNN [29] 75.0 46.5 66.7 0.40
BTPRED1 [25] 74.9 55.3 48.0 0.35
BTPRED [25, 35] 74.4 48.3 57.3 0.35
Chou-Fasman [10, 35] 65.2 37.6 63.5 0.26
Thornton [17, 35] 68.0 38.6 52.4 0.23
GORBTURN [18, 35] 70.5 39.3 37.3 0.19
1–4 & 2–3 correlation [19, 35] 59.1 32.4 61.9 0.17
Sequence coupled [20, 35] 53.3 32.4 72.8 0.17
  1. 1 Results reported on a different dataset with 300 chains.