A similar approach was demonstrated to work well for single-end reads with the tool BS Seeker, which was developed independently of Bismark (Chen et al., 2010). BS Seeker outperformed earlier generation BS-Seq mapping programs such as BSMAP, RMAP-bs or MAQ in terms of mapping efficiency, accuracy and required CPU time. Even though the principle of both tools is similar, Bismark offers a number of advantages over BS Seeker which are summarized in Table 1. For a test dataset [15 million reads taken from SRR020138 (Lister et al., 2009), trimmed to 50 bp, mapped to the human genome build NCBI36, one mismatch allowed], a direct comparison of the two tools returned a very similar number of alignments in a similar time scale [aligned reads/mapping efficiency/CPU time: 9 633 448/64.2%/42 min (Bismark); 9 664 184/64.4%/29 min (BS Seeker)]. Due to the way Bismark determines uniquely best alignments, it is less likely to report non-unique alignments; however, this comes at the cost of a slightly increased run time (for details see Supplementary Material). Table 1.Feature comparison of Bismark and BS SeekerFeatureBismarkBS SeekerBowtie instances
