StackTTCA: a stacking ensemble learning-based framework for accurate and high-throughput identification of tumor T cell antigens

Table 5 Performance comparison of StackTTCA and top five ML classifiers

Evaluation strategy	Method	ACC	Sn	Sp	MCC	AUC
Cross-validation	ADA-CTD	0.816	0.829	0.803	0.636	0.893
	RF-CTD	0.832	0.854	0.810	0.667	0.912
	ET-CTD	0.833	0.848	0.818	0.669	0.917
	LGBM-CTD	0.847	0.861	0.833	0.697	0.921
	XGB-CTD	0.848	0.852	0.843	0.698	0.920
	StackTTCA	0.879	0.896	0.861	0.760	0.935
Independent test	ADA-CTD	0.827	0.822	0.832	0.654	0.918
	RF-CTD	0.895	0.949	0.840	0.794	0.942
	ET-CTD	0.869	0.881	0.857	0.739	0.945
	LGBM-CTD	0.899	0.898	0.899	0.797	0.951
	XGB-CTD	0.903	0.915	0.891	0.806	0.946
	StackTTCA	0.932	0.958	0.908	0.866	0.962

ISSN: 1471-2105