Binning sequences using very sparse labels within a metagenome

BMC Bioinformatics

Table 1 Clustering performance of semi-supervised algorithms. Performance is measured by the Adjusted Rand Index (ARI) and Weighted F-measure (WF). Results for COP K-means are the best results in 100 runs with different initial k cluster centres.

	COP K		Constrained K		Seeded K		TSVM		S-GSOM-55
	ARI	WF	ARI	WF	ARI	WF	ARI	WF	ARI	WF
10Sp Set 1	0.84	0.94	0.84	0.94	0.84	0.93	0.25	0.59	0.85	0.95
10Sp Set 2	0.89	0.96	0.79	0.90	0.78	0.90	0.41	0.69	0.93	0.97
10Sp Set 3	0.58	0.83	0.85	0.93	0.84	0.93	0.27	0.62	0.83	0.93
20Sp Set 1	0.91	0.90	0.77	0.82	0.76	0.82	0.45	0.65	0.97	0.96
20Sp Set 2	0.76	0.82	0.70	0.79	0.67	0.79	0.43	0.62	0.83	0.89
20Sp Set 3	0.81	0.89	0.75	0.86	0.75	0.86	0.46	0.67	0.97	0.98
40Sp	0.58	0.76	0.71	0.85	0.68	0.84	0.24	0.56	0.83	0.91

ISSN: 1471-2105