Table 1.

Numerical summary of the false-assignment rates and fractions of false index pairs observed for the three different experiments no-CAP, SP-CAP and MP-CAP

	no-CAP	SP-CAP	MP-CAP
Total number of raw reads	34 241 955	48 546 372	34 684 183
Index pairs in raw data
Correct pairs (*) (%)	89.14	78.83	89.38
False pairs (%)	0.582	0.509	0.760
False index pairs after PF-filtering of raw reads
Total number of PF-filtered reads	27 466 817	37 586 292	27 220 161
False pairs (%)	0.523	0.387	0.691
False index pairs after quality score based filtering of index reads
Average index quality filter (~PF) (%)	0.059	0.192	0.423
Mininum index quality filter (~PF) (%)	0.060	0.177	0.422
Minimum index quality score of 15 (%)	0.060	0.138	0.428
False index pairs after quality score based filtering of template reads
Read quality filter on both reads (~PF) (%)	0.362	0.439	0.614
Read quality filter on the forward read (~PF) (%)	0.389	0.394	0.593
Read quality filter on the reverse read (~PF) (%)	0.298	–	–
Quantifying cross-contamination, mixed clusters and jumping PCR
False pairs due to contamination (%)	0.042	0.104	0.038
False pairs due to mixed clusters / jump. PCR (%)	0.018	0.034	0.390

~PF indicates values for a quality score cutoff that removes fewer raw reads than Illumina's Pass Filter (PF) flag.

*The fraction of correct index pairs is strongly affected by loading density. Denser loading in experiment SP-CAP led to a higher sequencing error rate and hence reduced the fraction of correct index pairs.