m⁶A Reader: Epitranscriptome Target Prediction and Functional Characterization of N ⁶-Methyladenosine (m⁶A) Readers

N ⁶-methyladenosine (m⁶A) is the most abundant post-transcriptional modification in mRNA, and regulates critical biological functions via m⁶A reader proteins that bind to m⁶A-containing transcripts. There exist multiple m⁶A reader proteins in the human genome, but their respective binding specificity and functional relevance under different biological contexts are not yet fully understood due to the limitation of experimental approaches. An in silico study was devised to unveil the target specificity and regulatory functions of different m⁶A readers. We established a support vector machine-based computational framework to predict the epitranscriptome-wide targets of six m⁶A reader proteins (YTHDF1-3, YTHDC1-2, and EIF3A) based on 58 genomic features as well as the conventional sequence-derived features. Our model achieved an average AUC of 0.981 and 0.893 under the full-transcript and mature mRNA model, respectively, marking a substantial improvement in accuracy compared to the sequence encoding schemes tested. Additionally, the distinct biological characteristics of each individual m⁶A reader were explored via the distribution, conservation, Gene Ontology enrichment, cellular components and molecular functions of their target m⁶A sites. A web server was constructed for predicting the putative binding readers of m⁶A sites to serve the research community, and is freely accessible at: http://m6areader.rnamd.com.