Figure 1. Fewer clusters are identified by the Illumina purity filter in low-diversity samples.

(a) The total number of sequences per lane passing the purity filter step for samples with no bias, or one or two initially biased sequence (IBS) libraries. Data from three representative flow cell lanes are shown for each. (b) Current cluster-calling algorithms can discern clusters even at extremely high densities if the sequence composition during the first four cycles is unbiased. In low-diversity samples clusters with the same initially biased sequence in very close proximity may erroneously be called as a single cluster and may ultimately be removed by the purity filter.