BatMeth: improved mapper for bisulfite sequencing reads on DNA methylation

DNA methylation modifies the nucleotide cytosine by the addition of methyl groups to its C5 carbon residue by DNA methyltransferases [1]. This modification can be inherited through cell division and it plays an important role in many biological processes, such as heterochromatin and transcriptional silencing [2,3], imprinting genes [4], inactivating the × chromosome [5] and silencing of repetitive DNA components in healthy and diseased (including cancerous) cells [6,7]. Methylation analysis can also be used to diagnose pre-natal Down's syndrome [8]. Thus, the genome-wide methylation profiles of different tissues are important to understand the complex nature and effects of DNA methylation.In the past decade, quantum leaps have been made in the development of sequencing technologies by vendors such as Illumina-Solexa and Applied BioSystems (AB)-SOLiD. These can generate millions of short reads at a lower cost compared to traditional Sanger methods [9-13]. Bisulfite (BS) treatment converts unmethylated cytosines (Cs) to uracils (which are then amplified by PCR as thymine (T)) without affecting the other nucleotide bases and methylated cytosines [14]. Next-generation sequencing coupled with bisulfite treatment enables us to produce a methylome of a genome at single base resolution and low cost.One important step in calling methylation of a genome is to map bisulfite reads. Mapping of bisulfite reads is different from that of ChIP-Seq and RNA-Seq data since the non-methylated Cs are converted to Ts by bisulfite treatment and subsequent PCR. The bisulfite reads are difficult to map to the reference genome due to the high number of mismatches between the converted Ts and the original Cs. For mapping Illumina bisulfite reads, the pioneering published methods are BSMAP [15] and RMAP [16]. BSMAP aligns a bisulfite read to the reference genome by first enumerating all C-to-T combinations within a user-defined length k seed of the reads; then, through hashing, BSMAP aligns the