BMC Bioinformatics

Table 4 Comparison of feature-based results based on the macro-F1 score on various window sizes. For each classification model, the best results are tabulated in bold

From: Machine learning-based approaches for ubiquitination site prediction in human proteins

Window size	Feature	Method
Window size	Feature	KNN	XGBoost	RF	SVM	DNN
9	PSSM	0.441	0.451	0.439	0.450	0.520
	AAC	0.507	0.437	0.435	0.416	0.489
	DPC	0.491	0.435	0.435	0.506	0.518
	Physicochemical	0.422	0.460	0.429	0.46	0.516
15	PSSM	0.454	0.468	0.440	0.441	0.526
	AAC	0.446	0.462	0.452	0.435	0.503
	DPC	0.458	0.462	0.452	0.508	0.523
	Physicochemical	0.423	0.466	0.430	0.467	0.533
21	PSSM	0.440	0.452	0.440	0.419	0.523
	AAC	0.447	0.450	0.459	0.441	0.500
	DPC	0.438	0.457	0.450	0.488	0.523
	Physicochemical	0.425	0.462	0.431	0.445	0.537

Back to article page

ISSN: 1471-2105

Contact us

General enquiries: journalsubmissions@springernature.com