Identification of Hot Spots in Protein Structures Using Gaussian Network Model and Gaussian Naive Bayes

Table 1

List of top 20 measures based on tenfold cross validations of Gaussian Naive Bayes when using single th highest mode () inputted into the feature vector, where cutoff means the distance threshold for GNM computation that varies from 6.0 to 8.0 with step size of 0.1 and sw represents the size of the sliding window for the central residue that ranges from 1 to 21 with step size of 2.

Top	Cutoff	i	sw	sen	spe	pre	acc	measure
1	7.3	8	3	0.1930	0.9436	0.1250	0.9136	0.1517
2	7.1	8	9	0.2515	0.9095	0.1039	0.8831	0.1470
3	7.1	8	7	0.2456	0.9119	0.1042	0.8852	0.1463
4	7.1	8	5	0.2164	0.9263	0.1091	0.8979	0.1451
5	7.1	8	3	0.1696	0.9473	0.1184	0.9162	0.1394
6	7.3	8	5	0.1871	0.9354	0.1077	0.9054	0.1368
7	8.0	3	5	0.1930	0.9310	0.1044	0.9014	0.1355
8	7.3	8	7	0.2164	0.9163	0.0974	0.8883	0.1343
9	7.1	8	13	0.2281	0.9090	0.0947	0.8817	0.1338
10	7.1	8	11	0.2281	0.9071	0.0929	0.8799	0.1320
11	7.0	19	17	0.2456	0.8963	0.0899	0.8703	0.1317
12	6.7	13	3	0.1345	0.9619	0.1285	0.9288	0.1314
13	7.8	3	3	0.1520	0.9507	0.1140	0.9187	0.1303
14	7.0	14	21	0.2339	0.901	0.0897	0.8742	0.1297
15	7.0	19	19	0.2456	0.8934	0.0877	0.8674	0.1292
16	7.1	8	15	0.2281	0.9039	0.0901	0.8768	0.1291
17	7.0	4	7	0.2281	0.9022	0.0886	0.8752	0.1277
18	6.6	6	3	0.1520	0.9480	0.1088	0.9162	0.1268
19	6.9	15	21	0.2222	0.9046	0.0886	0.8773	0.1267
20	7.2	14	13	0.2456	0.8897	0.0850	0.8639	0.1263