Table 2.

Performance of SVM models developed using PSSM profile of proteins at different window lengths.

Window size	Kernel parameters	Thr*	Sen (%)	Spe (%)	Acc (%)	MCC
3	t 2 g 1.0 j 1 c 10	0	83.26	82.61	82.93	0.66
5	t 2 g 1.0 j 1 c 10	0	82.59	87.51	85.05	0.7
7	t 2 g 0.1 j 1 c 10	0	82.39	84.67	83.53	0.67
9	t 2 g 0.1 j 1 c 10	0	84.18	86.13	85.16	0.7
11	t 2 g 0.1 j 1 c 10	0	85.28	86.25	85.77	0.72
13	t 2 g 0.1 j 1 c 10	0	85.7	86.52	86.11	0.72
15	t 2 g 0.1 j 1 c 10	0	85.36	86.31	85.84	0.72
17	t 2 g 0.1 j 1 c 10	0	83.69	87.99	85.84	0.72
19	t 2 g 0.1 j 1 c 10	0	86.13	88.37	87.25	0.75
21	t 2 g 0.1 j 1 c 10	0	85.52	87.33	86.43	0.73

SVM models were trained and tested on a dataset having equal number of positive and negative data. Bold font shows the performance and parameters of selected SVM model.