As the strand identity of a bisulfite read is a priori unknown, our bisulfite mapping tool Bismark aims to find a unique alignment by running four alignment processes simultaneously. First, bisulfite reads are transformed into a C-to-T and G-to-A version (equivalent to a C-to-T conversion on the reverse strand). Then, each of them is aligned to equivalently pre-converted forms of the reference genome using four parallel instances of the short read aligner Bowtie (Fig. 1A). This read mapping enables Bismark to uniquely determine the strand origin of a bisulfite read. Consequently, Bismark can handle BS-Seq data from both directional and non-directional libraries. Since residual cytosines in the sequencing read are converted in silico into a fully bisulfite-converted form before the alignment takes place, mapping performed in this manner handles partial methylation accurately and in an unbiased manner.
