ASMS: finding allele specific methylation in human genomes without phasing

Abstract

MotivationAllele-specific methylation (ASM) refers to differential DNA methylation patterns between two alleles at a given locus. This phenomenon is often driven by genetic variants, such as single nucleotide variants (SNVs), which influence methylation in cis by affecting transcription factor or methylation regulator binding, leading to allele-specific differences. Understanding ASM is critical for elucidating gene regulation, as it impacts gene expression and contributes to normal biological variation as well as disease processes, including cancer [1] and autoimmune disorders [2],[3]. Another key driver of ASM is genomic imprinting, an epigenetic mechanism in which gene expression is regulated in a parent-of-origin-specific manner. Imprinted regions, marked during gametogenesis and maintained through cell divisions, are essential for growth, development, and metabolism. Dysregulation of imprinting is associated with developmental and metabolic disorders, such as Prader-Willi and Angelman syndromes, and certain cancers. Detecting ASM across the genome remains challenging due to its tissue- and cell-specific nature and the technical difficulty of phasing reads to assign methylation patterns to specific alleles. Current ASM detection pipelines (e.g., [4])) often require phasing via genetic variants, a computationally intensive process that is limited in regions with low heterozygosity. ResultsTo address these limitations, we developed asms (Allele-Specific Methylation Scanner), a tool designed to detect ASM directly from methylation data without the need for prior phasing. asms offers a faster and complementary approach to uncovering the regulatory effects of ASM, particularly in genomic regions where genetic variants or imprinting play a critical role. For demonstration purposes we leverage the fact that reads generated by Oxford Nanopore (ONT) technology measure sequence and methylation status at once, but the same software can be used with other sequencing technologies. asms can check thousands of loci in a short time. The initial list of loci to examine can be given by the user, or generated by asms through a genomic scan. The asms cluster subcommand separates the reads based on methylation. If phasing results are available, asms can use them to verify whether distinct alleles correspond to distinct base modifications patterns. We benchmark our software using publicly available Ashkenazi trio data [5]. Implementation and availabilityasms is implemented in rust and python. The software is available at https://github.com/ecmra/asms.