论文部分内容阅读
Transcriptome N6-methyladenine (m6A) methylation affects the regulation of various biological processes such as RNA stability and mRNA translation.Altered transcriptome methylation is known to contribute to etiology of human diseases.Detecting the differentially-methylated m6A sites under different conditions from methylated RNA immunoprecipitation sequencing (MeRIP-Seq) data is essential to understand the mechanism of m6A in disease.Identification of differential m6A site from MeRIP-Seq requires comparing IP/input read enrichments from two conditions.However, since these samples are usually sequenced at different library size, new differential analysis approach that can properly normalize different library size is needed.We propose here bltest, a new likelihood ratio test based on binomial models for identifying differentially-methylated m6A sites.Compensation of sequencing depth is implicitly included in the model of bltest and thus no separate normalization is needed.Simulation results show that bltest consistently outperforms exomePeak, the existing alternative, especially when the sequencing depths of MeRIP-Seq samples are considerably different.bltest was applied to 3 real MeRIP-Seq datasets of different conditions and revealed highly context-dependent landscape of global m6A.The bltest algorithm has been integrated into the open source exomePeak R/Bioconductor package and is freely available at http://www.bioconductor.org.