Table 2.

Overview table comparing error rates of the samples treated with UNOISE (after USEARCH preprocessing) and those without applying a denoising algorithm, after applying Pre-cluster or after applying IPED (after mothur preprocessing). Due to the difference preprocessing steps applied in USEARCH and mothur, the amount of reads removed differ, where around 53 % and 39 % of reads are removed in respective order

Variable region	Sample ID			Error rates
		USEARCH + UNOISE	Mothur (make.contigs)	Mothur + Pre-cluster	Mothur + IPED
V34	130403	0.0003	0.0026	0.0013	0.0002
	130417	0.0004	0.0023	0.0010	0.0003
	130422	0.0008	0.0028	0.0017	0.0008
	M1	0.0006	0.00149	0.0007	0.0004
	M2	0.0007	0.00150	0.0008	0.0006
	M3	0.0005	0.00140	0.0007	0.0005
V4	130403	0.00011	0.00056	0.00013	0.00010
	130417	0.00009	0.00051	0.00010	0.00008
	130422	0.00009	0.00049	0.00010	0.00008
	v4.I.1	0.00002	0.00061	0.00008	0.00004
	v4.I.05	0.00002	0.00068	0.00010	0.00004
V45	130403	0.0030	0.0084	0.0055	0.0022
	130417	0.0029	0.0069	0.0041	0.0020
	130422	0.0026	0.0060	0.0033	0.0016
	v4.v5.I.1	0.0082	0.0066	0.0061	0.0041
	v4.v5.I.11	0.0084	0.0033	0.0034	0.0031
Average	All samples	0.0018	0.0029	0.0018	0.0010