Sequence-based bacterial small RNAs prediction using ensemble learning strategies

BMC Bioinformatics

Table 3 Performances of individual feature-based models constructed by RF and SVM on the balanced dataset

Index	Feature	AUC		ACC		SN		SP
Index	Feature	RF	SVM	RF	SVM	RF	SVM	RF	SVM
F1	1-spectrum profile	0.682	0.657	0.560	0.512	0.912	0.985	0.209	0.039
F2	2-spectrum profile	0.829	0.821	0.756	0.749	0.792	0.788	0.720	0.711
F3	3-spectrum profile	0.909	0.874	0.834	0.800	0.863	0.835	0.805	0.765
F4	4-spectrum profile	0.923	0.909	0.860	0.840	0.873	0.866	0.846	0.814
F5	5-spectrum profile	0.912	0.896	0.842	0.822	0.847	0.874	0.838	0.770
F6	(3, m)-mismatch profile	0.769	0.795	0.679	0.717	0.807	0.812	0.552	0.622
F7	(4, m)-mismatch profile	0.880	0.885	0.797	0.816	0.814	0.843	0.780	0.789
F8	(5, m)-mismatch profile	0.913	0.907	0.835	0.832	0.848	0.882	0.822	0.782
F9	1-RevcKmer	0.632	0.655	0.516	0.542	0.972	0.935	0.060	0.150
F10	2-RevcKmer	0.842	0.804	0.765	0.726	0.828	0.817	0.702	0.636
F11	3-RevcKmer	0.924	0.868	0.855	0.791	0.848	0.831	0.863	0.750
F12	4-RevcKmer	0.938	0.894	0.880	0.818	0.880	0.869	0.880	0.768
F13	5-RevcKmer	0.937	0.906	0.874	0.829	0.859	0.856	0.889	0.802
F14	PCPseDNC	0.895	0.905	0.827	0.828	0.850	0.868	0.803	0.787
F15	PCPseTNC	0.931	0.922	0.862	0.857	0.856	0.848	0.868	0.865
F16	SCPseDNC	0.902	0.888	0.825	0.811	0.841	0.810	0.809	0.811
F17	SCPseTNC	0.905	0.910	0.825	0.840	0.854	0.841	0.795	0.839

ISSN: 1471-2105