Grant: Travel Awards - Institute Community Support

Background & Objective: Economical inference on DNA methylation status in genomic regions of interest has been made possible by targeted next-generation bisulfite sequencing (NGBS). Selecting a suitable aligner to map the short reads generated from targeted NGBS back to the reference genome plays a key role in the accurate determination of DNA methylation profiles. We recently conducted a small pilot study to evaluate the performance and feasibility of three efficient bisulfite sequence aligners - Novoalign, GemBS, and Bismark - for a larger project consisting of 1,000 or more subjects. To date, evaluation and comparisons of these methods have been limited to whole genome sequencing experiments. We wished to determine the performance of these three methods using targeted NGBS approaches. Our pilot experiment evaluated each aligner using 24 samples, with the following hardware specifications: processing speed of 2.67 GHz and maximum memory of 192 GB for each compute node. Methods: The performance of bisulfite sequencing aligners is determined by three parameters mapping efficiency, maximum memory usage, and mapping time. Results: Novoalign has the highest mapping efficiency (89%), as determined by the number of reads mapped to unique locations in the reference genome, outperforming GemBS (76%) and Bismark (72%) by 13% and 17% respectively. GemBS is the fastest aligner, with an average mapping time of just 32 minutes per sample, while Novoalign and Bismark need 25.7 and 9.1 hours respectively. In terms of memory usage, Novoalign and Bismark consume less than 4.5 GB of memory, and GemBS requires 6.6 GB, more than 1.5 times the memory allocation of Novoalign and Bismark. Conclusion: GemBS is the most suitable application for large-scale DNA methylation analysis projects if its compromised mapping efficiency and large memory requirement are acceptable to researchers. Novoalign is an option for projects whose aims at higher mapping efficiency.