Table 3.

Feature comparison and selection based on a subset with 1000 data points

Feature	Sensitivity	Specificity	Precision	F1
AVE_ESP1_0.5	0.31	0.94	0.54	0.40
(Feature 39)
Local Amino Acid	0.24	0.85	0.27	0.26
Microenvironment
(Features 133–152)
PSSM[i − 5, i + 5]	0.32	0.93	0.52	0.40
(Features 221–440)

Training results for best electrostatic-based feature (Table 1), best PSSM-based feature group (Table 2), and best residue microenvironment feature group are compared.