ARYANA-BS: Context-Aware Alignment of Bisulfite-Sequencing Reads

MotivationDNA methylation plays a crucial role in biological processes, including imprinting, development, inflammation, and several disorders, such as cancer. Bisulfite sequencing (BS) is the gold standard for single-base resolution in measuring DNA methylation. This process involves treating genomic DNA with sodium bisulfite, followed by polymerase chain reaction (PCR), converting unmethylated cytosines to thymines (C to T) and guanines to adenines (G to A). However, aligning reads obtained through next-generation sequencing (NGS) of the converted DNA is challenging due to the high number of mismatches caused by these conversions. Various BS aligners aim to simplify BS read alignment to conventional DNA alignment by modifying the reference genome and/or reads. Methods include three-letter alignment and wild-card alignment, each with its limitations. This work introduces a novel BS aligner, ARYANA-BS, which departs from conventional DNA aligners by considering base alterations in BS reads within its alignment engine. Leveraging well-established DNA methylation patterns in different genomic contexts, ARYANA-BS generates five indexes from the reference, aligns each read to all indexes, and selects the hit with the minimum penalty. To further enhance alignment accuracy, an optional EM step has been developed, incorporating methylation probability information in the decision-making process for the index with the minimum penalty for each read. The presented approach seeks to improve the accuracy of BS read alignment by accounting for the intricacies of DNA methylation patterns in diverse genomic contexts. ResultsOur experimental results, based on both simulated and real data, demonstrate that ARYANA-BS achieves state-of-the-art accuracy while maintaining competitive speed and memory usage. AvailabilityThe source code of ARYANA and ARYANA-BS, the read simulator for both normal and bisulfite-treated reads, SAM file analyzer which is used for post processing of the alignment penalties, and test procedures for benchmarking different aligners using simulated and real data, are publicly available in https://github.com/hnikaein/aryana. Contactsharifi@sharif.edu Supplementary informationSupplementary data are available at Journal Name online.