A modular kernel approach for integrative analysis of protein domain boundaries

BMC Genomics

Table 1 Predictive performance of machine-learning models

Models	TN_f	TP_f	Sensitivity (Sn)	Specificity (Sp)	Correlation- Coefficient (Cc)	Accuracy (Ac)
HME_HE	0.77 ± 0.015	0.79 ± 0.026	0.78 ± 0.002	0.78 ± 0.012	0.56 ± 0.016	0.78 ± 0.015
HME_PSSM	0.74 ± 0.019	0.74 ± 0.018	0.75 ± 0.010	0.73 ± 0.045	0.48 ± 0.023	0.74 ± 0.016
SVM_HE	0.71 ± 0.008	0.73 ± 0.010	0.70 ± 0.003	0.74 ± 0.017	0.44 ± 0.011	0.72 ± 0.020
SVM_PSSM	0.71 ± 0.004	0.67 ± 0.008	0.65 ± 0.012	0.72 ± 0.006	0.37 ± 0.007	0.69 ± 0.003
MLP_HE	0.69 ± 0.009	0.72 ± 0.012	0.61 ± 0.027	0.75 ± 0.019	0.40 ± 0.013	0.70 ± 0.025
MLP_PSSM	0.67 ± 0.017	0.71 ± 0.032	0.61 ± 0.013	0.76 ± 0.027	0.37 ± 0.022	0.68 ± 0.011

Mean testing data ± standard deviation obtained using ANOVA test using optimal settings for each model.

ISSN: 1471-2164