Differential analysis of RNA methylome with improved spatial resolution

Recent development of MeRIP-Seq enabled the global unbiased profiling of transcriptome-wide N6-Adenosine. With this technique, it is now possible to detect the RNA methylation sites under a specific condition or the differential methylation sites between two experimental conditions. However, as an affinity-based approach, MeRIP-Seq has yet provided base-pair resolution. A single methylation site reported by MeRIP-Seq data may actually contain one or a few methylated RNA residuals, which cannot be differentiated by existing differential analysis methods when the entire RNA methylation site is treated as a single feature. Within this paper, we propose a new approach `RHHMM' that combines Fisher's exact test and hidden Markov model (HMM) for the detection of differential methylation regions (DMRs) with improved spatial resolution. The results on both simulated and real data demonstrated that, with HMM incorporating local spatial dependency, it is possible to detect differential methylation sites with improved spatial resolution on affinity-based sequencing approach such as MeRIP-Seq. The proposed method is freely available as an open source R package.