Deep learning models for bacteria taxonomic classification of metagenomic data

BMC Bioinformatics

Table 4 Comparison among classification performances of CNN, DBN and RDP algorithms at varying of k-mer size. for both SG and AMP datasets

Evaluation of short-reads classification at genus level
Dataset	Algorithm	k	Accuracy		Precision		Recall		F1
			mean %	std	mean %	std	mean %	std	mean %	std
AMP	CNN	3	51.01	0.005	51.40	0.005	50.90	0.005	50.84	0.015
		4	77.69	0.004	77.91	0.005	77.69	0.005	77.57	0.014
		5	88.13	0.005	88.38	0.005	88.07	0.006	88.98	0.014
		6	90.92	0.005	91.14	0.005	90.91	0.005	90.82	0.009
		7	91.33	0.004	91.57	0.004	91.32	0.004	91.18	0.015
	DBN	3	56.69	0.013	57.88	0.011	56.62	0.013	55.56	0.013
		4	85.10	0.004	85.47	0.005	85.08	0.004	84.53	0.008
		5	89.82	0.003	90.12	0.004	89.82	0.003	89.63	0.004
		6	90.55	0.005	90.73	0.005	90.53	0.005	90.45	0.005
		7	91.37	0.005	91.62	0.005	91.37	0.005	91.26	0.005
	RDP	-	83.84	0.007	84.42	0.007	83.57	0.007	83.65	0.007
SG	CNN	3	17.02	0.018	17.32	0.013	16.53	0.015	16.69	0.006
		4	32.98	0.015	33.42	0.012	32.59	0.013	32.65	0.005
		5	59.80	0.015	60.34	0.014	59.41	0.015	59.31	0.005
		6	80.77	0.009	81.10	0.010	80.41	0.009	80.33	0.005
		7	85.50	0.014	85.70	0.014	85.20	0.014	85.11	0.005
	DBN	3	17.75	0.009	19.80	0.010	17.50	0.009	16.32	0.010
		4	54.11	0.007	55.62	0.007	53.67	0.007	53.17	0.007
		5	71.44	0.007	72.45	0.009	71.07	0.007	70.99	0.008
		6	77.85	0.007	78.36	0.008	77.53	0.008	77.47	0.008
		7	81.27	0.002	81.87	0.004	80.92	0.003	80.94	0.002
	RDP	-	80.38	0.009	80.83	0.008	80.18	0.008	80.09	0.009

ISSN: 1471-2105