m6AConquer: a Data Resource for Unified Quantification and Integration of m6A Detection Techniques

N6-methyladenosine (m6A) is the most prevalent RNA modification in mammalian cells and the most extensively studied epitranscriptomic mark. More than 10 m6A detection techniques have been proposed to measure m6A stoichiometry at either the site or peak level. However, these detection techniques are processed through heterogeneous pipelines, using different computational filters and reference features, leading to difficulties in fully harnessing data integration and analysis across orthogonal m6A detection techniques. Our m6AConquer (Consistent Quantification of External m6A RNA Modification Data) tackles this challenge by establishing a consistent multi-omics data-sharing standard, summarizing quantitative m6A data from 10 detection techniques using a unified reference feature set. Furthermore, we standardize site calling and m6A count matrix normalization procedures across platforms through a computational framework that accounts for over-dispersion in m6A levels. Available m6A detection techniques can be categorized into four types: antibody-assisted, chemical-assisted, enzyme-assisted, and direct-RNA sequencing. We leverage this categorization to develop a reproducibility-based integration framework that enables the reliable detection of high-confidence m6A sites confirmed across orthogonal techniques. Empirical evaluations report that both the site-calling and the integration framework outperform common alternatives, enhancing biological relevance. We apply interpretable machine learning models on our integrated high-confidence sites, and the results consistently identify proximity to intron-exon junctions as the driving predictor of m6A site coordinates across different techniques, demonstrating the high quality of the data curated in m6AConquer. m6AConquer webserver is freely accessible under: http://rnamd.org/m6aconquer.