Prediction of bacterial type IV secreted effectors by C-terminal features

BMC Genomics

Table 1 Performance of different models classifying T4S effectors and non-effectors

Features	Model	Sn(%) vs. Sp(%)	A(%)	*AUC*	*MCC*
Seq_Aac	SVM	50.57 vs. 93.86	79.43	0.8212	0.5146
Seq_bAac	SVM	44.57 vs. 96.29	79.05	0.8311	0.5088
Seq_Aac, bAac	SVM	46.00 vs. 96.14	79.43	0.8343	0.5182
Seq_Sig	SVM	50.57 vs. 93.86	79.43	0.8500	0.5146
Motif	-	50.43 vs. 88.18	75.60	-	0.4222
Seq_Aac, Sse, Acc	SVM	69.71 vs. 91.14	84.00	0.8742	0.6313
Pos_Aac_SPB	SVM	61.71 vs. 92.14	82.00	0.8538	0.5802
Pos_Aac _SPB + Seq_Aac	SVM	78.86 vs. 93.29	88.48	0.9362	0.7369
Pos_Aac_BPB	BPB-SVM	79.14 vs. 94.43	89.33	0.9559	0.7561
Pos_Aac, Sse, Acc	BPB-SVM	89.14 vs. 97.14	94.57	0.9883	0.8770

Note: The RBF kernel function was used for all the models except ‘Motif’. The performance was evaluated according to 5-fold cross validation results.

ISSN: 1471-2164