From Detection to Prediction: Advances in m6A Methylation Analysis Through Machine Learning and Deep Learning with Implications in Cancer

Ruoting Jin, Quan Zou, Ximei Luo

N6-methyladenosine (m6A) represents the most common and thoroughly investigated RNA modification and exerts essential functions in regulating gene expression through influencing the RNA stability, the translation efficiency, alternative splicing, and nuclear export processes. The rapid development of high-throughput sequencing approaches, including miCLIP and MeRIP-seq, has profoundly transformed epitranscriptomics research. These techniques facilitate the detailed transcriptome-wide profiling of m6A modifications, shedding light on their crucial roles in diverse biological pathways. This review comprehensively examines the identification, mechanisms of regulation, and functional consequences of m6A modifications. It emphasizes their critical roles in physiological contexts, encompassing immune function, neuronal development, and the differentiation of stem cells. Additionally, the review discusses the contributions of m6A dysregulation to pathological conditions, including cancer, neurodegenerative diseases, and disorders of metabolism. We also discuss the development and application of machine-learning algorithms for m6A site prediction, emphasizing the integration of sequence-based, structural, and evolutionary conservation features to enhance the predictive accuracy. Furthermore, the potential of applying the findings from m6A research in precision medicine and drug development is examined. By synthesizing the current knowledge and emerging trends, this review aims to provide a comprehensive understanding of m6A biology and its translational potential, offering new perspectives for future research and therapeutic innovation.