Evaluation of genomic high-throughput sequencing data generated on Illumina HiSeq and Genome Analyzer systems

Table 2 Expected error rates after filtering

	Expected error rate^a(percentage of bases discarded)
Filter	Bv + PhiX	At + PhiX	PhiX-GAIIx
No filter	7.093 (0.0%)	10.619 (0.0%)	16.434 (0.0%)
ChF	2.583 (10.2%)	4.819 (12.7%)	7.360 (17.8%)
B-tail	0.163 (11.0%)	0.205 (16.6%)	0.229 (25.8%)
N	5.943 (2.5%)	9.688 (2.3%)	8.601 (15.6%)
C33	1.521 (14.3%)	2.907 (17.7%)	5.207 (21.7%)
A30	1.802 (12.7%)	3.586 (15.5%)	5.457 (21.3%)
B-tail + N	0.141 (11.4%)	0.187 (17.0%)	0.206 (26.8%)
B-tail + ChF	0.118 (13.8%)	0.161 (19.2%)	0.204 (27.2%)
B-tail + ChF + C33	0.083 (16.9%)	0.127 (22.1%)	0.174 (28.3%)
B-tail + ChF + A30	0.093 (15.8%)	0.139 (21.0%)	0.176 (28.2%)
B-tail + ChF + N + C33	0.077 (17.1%)	0.125 (22.2%)	0.168 (28.9%)
B-tail + ChF + N + A30	0.085 (16.0%)	0.136 (21.1%)	0.171 (28.6%)

^aExpected error rate: average error probability of each base assigned by Illumina as quality scores. No filter, raw data prior to filtering; ChF, Illumina chastity filter; B-tail, B-tail trimming; N, removal of reads with at least one uncalled base; C33, removal of reads having less than two-thirds of bases with Q ≥ 30 within the first half of the read; A30, removal of reads that have an average Q-score < 30 in the first 30% of the read. Note that some quality filters discard partially the same reads (for instance, the chastity filter and the removal of reads containing at least one uncalled base).

ISSN: 1474-760X