Comparative analysis and prediction of nucleosome positioning using integrative feature representation and machine learning algorithms

Table 14 Comparison of our predictors with other advanced models via 20-fold cross-validation for D. melanogaster

Method	Feature	ACC	Sn	Sp	MCC	AUC
LeNup [25]		0.8847	0.8974	0.8713	0.7828	0.9401
3LS [22]		0.8341	0.8407	0.8274	0.6682	0.9147
SVM	FCGR	0.8117	0.7841	0.8396	0.6251	0.8782
	FCGR + DAC	0.8094	0.7876	0.8316	0.6201	0.8783
	FCGR + TAC	0.8118	0.8073	0.8163	0.6252	0.8863
	FCGR + DACC	0.7866	0.7997	0.7733	0.5738	0.8384
	FCGR + TACC	0.7767	0.7128	0.8418	0.5593	0.8412
	FCGR + PseDNC	0.8095	0.7992	0.8199	0.6209	0.8843
	FCGR + PseTNC	0.8108	0.8034	0.8183	0.6234	0.8848
	All features	0.7602	0.7014	0.8200	0.5255	0.8059
ELM	FCGR	0.7912	0.7651	0.8204	0.5842	0.8601
	FCGR + DAC	0.7924	0.7752	0.8098	0.5862	0.8689
	FCGR + TAC	0.7932	0.7776	0.8091	0.5877	0.8619
	FCGR + DACC	0.7793	0.7710	0.7877	0.5599	0.8537
	FCGR + TACC	0.7697	0.7686	0.7709	0.5403	0.8456
	FCGR + PseDNC	0.7910	0.7659	0.8165	0.5837	0.8648
	FCGR + PseTNC	0.7691	0.7455	0.7930	0.5395	0.8433
	All features	0.7878	0.7859	0.7899	0.5763	0.8637
XGBoost	FCGR	0.8037	0.7821	0.8257	0.6088	0.8771
	FCGR + DAC	0.7891	0.7741	0.8042	0.5791	0.8648
	FCGR + TAC	0.7948	0.7762	0.8137	0.5910	0.8690
	FCGR + DACC	0.7814	0.7790	0.7839	0.5634	0.8540
	FCGR + TACC	0.7706	0.7648	0.7765	0.5417	0.8508
	FCGR + PseDNC	0.8010	0.7786	0.8239	0.6036	0.8728
	FCGR + PseTNC	0.8074	0.7958	0.8193	0.6165	0.8831
	All features	0.7979	0.7745	0.8218	0.5972	0.8739
MLP	FCGR	0.8127	0.8003	0.8253	0.6272	0.8893
CNN	FCGR	0.8116	0.8036	0.8198	0.6252	0.8854

ISSN: 1471-2105