Protein-protein interface hot spots prediction based on a hybrid feature selection strategy

BMC Bioinformatics

Table 4 Features selected and the corresponding cross-validation performance in PSFS process

Round	Features identified	Accuracy	Recall	Precision	F-measure
1	CNSV_REL1, BsRASA	0.766	0.790	0.681	0.731
	CNSV_REL1, BpRASA	0.779	0.710	0.733	0.721
	CNSV_REL1, BtRASA	0.753	0.694	0.694	0.694
2	CNSV_REL1, BsRASA, UsASA	0.818	0.774	0.774	0.774
	CNSV_REL1, BtRASA, UpASA	0.799	0.790	0.731	0.760
	CNSV_REL1, BsRASA, UpASA	0.799	0.774	0.739	0.756
3	CNSV_REL1, BsRASA, UpASA, BtRASA	0.818	0.807	0.758	0.781
	CNSV_REL1, BtRASA, UpASA, DtASA^1/2	0.812	0.807	0.746	0.775
	CNSV_REL1, BsRASA, UpASA, BpRASA	0.812	0.807	0.746	0.775
4	CNSV_REL1, BtRASA, UpASA, DtASA^1/2, BsRASA	0.825	0.838	0.754	0.794
	CNSV_REL1, BsRASA, UpASA, BtRASA, DpRASA^1/2	0.818	0.823	0.750	0.785
	CNSV_REL1, BsRASA, UpASA, BpRASA, DtASA^1/2	0.825	0.823	0.761	0.791
5	CNSV_REL1, BtRASA, UpASA, DtASA^1/2, BsRASA, BtmDI	0.831	0.839	0.765	0.800
	CNSV_REL1, BtRASA, UpASA, DtASA^1/2, BsRASA, B factor	0.831	0.839	0.765	0.800
	CNSV_REL1, BsRASA, UpASA, BpRASA, DtASA^1/2, BtmDI	0.825	0.823	0.761	0.791
6	CNSV_REL1, BtRASA, UpASA, DtASA^1/2, BsRASA, B factor, BtmDI	0.831	0.839	0.765	0.800
	CNSV_REL1, BtRASA, UpASA, DtASA^1/2, BsRASA, BtmDI, Hdrpi	0.831	0.839	0.765	0.800
	CNSV_REL1, BtRASA, UpASA, DtASA^1/2, BsRASA, B factor, BminPI	0.831	0.839	0.765	0.800

ISSN: 1471-2105