Characterizing and measuring bias in sequence data

Table 3 Percentage of undercovered microbial genome given 100× coverage

Data set			Relative coverage thresholds (% of genome)
Sample	#	Sequencing platform	= 0	≤0.1	≤0.25	≤0.5
P. falciparum 3D7	1	Illumina MiSeq	0.010	0.18	3.7	24
	2	Ion Torrent PGM	2.6	14	22	33
	3	Pacific Riosciences RS	0.012	0.13	0.33	2.7
	1+2	Illumina + Ion Torrent	0.0096	1.1	12	30
	2+3	Ion Torrent + Pac Bio	0.0062	0.097	1.6	17
	3+1	Pac Bio + Illumina	0.0051	0.040	0.33	7.9
E. coli K12 MG1655	4	Illumina MiSeq	0.00022	0.0019	0.019	0.54
	5	Ion Torrent PGM	0.00047	0.013	0.046	0.27
	6	Pacific Biosciences RS	0	0.00075	0.030	0.36
	4+5	Illumina + Ion Torrent	0	0.0012	0.0053	0.075
	5+6	Ion Torrent + Pac Bio	0	0.00037	0.0018	0.054
	6+4	Pac Bio + Illumina	0	0.00026	0.0012	0.061
R. sphaeroides 2.4.1	7	Illumina MiSeq	0.00094	0.045	0.39	2.7
	8	Ion Torrent PGM	0.88	19	36	47
	9	Pacific Biosciences RS	0.000048	0.0021	0.0071	0.067
	7+8	Illumina + Ion Torrent	0.0038	0.23	1.8	19
	8+9	Ion Torrent + Pac Bio	0.000024	0.00058	0.14	16
	9+7	Pac Bio + Illumina	0.00012	0.0018	0.017	0.49

The percentage of each microbial genome left completely uncovered as well as ten-fold, four-fold, and two-fold undercovered given 100× (randomly downsampled) coverage of Illumina MiSeq, Ion Torrent PGM, and Pacific Biosciences RS data (data sets in italics; Table 2). Also included are the same statistics from data sets that combine 50× coverage from each pair of sequencing technologies.

ISSN: 1474-760X