m6AGE

m6AGE predicts N6-methyladenosine (m6A) sites by integrating sequence-derived features and graph-embedding-based geometrical information to identify RNA modification sites for downstream analyses of gene regulation.

Key Features:

• Sequence-Derived Features: Utilizes characteristics extracted from RNA sequences to capture biological signals associated with m6A modifications.
• Graph Embedding-Based Geometrical Information: Incorporates graph embeddings to represent geometrical and structural relationships among sequence positions.
• Dual-Feature Integration: Integrates sequence-derived features with graph-embedding information, reported as the first integration of these two feature types for m6A prediction.
• Benchmark Evaluation: Evaluated on four public datasets from three species with reported improvements of Accuracy +1.34%, Matthew's Correlation Coefficient +0.0227, Specificity +5.63%, and AUC +0.0081 on the A101 dataset.

Scientific Applications:

• m6A site identification: Predicts locations of N6-methyladenosine modifications in RNA sequences.
• Functional studies of RNA modifications: Supports investigation of m6A roles in mRNA splicing, export, and stability.
• Comparative epitranscriptomics: Enables cross-species comparison of m6A site distribution using predictions across multiple species.

Methodology:

Combines sequence-derived feature extraction with graph-embedding-based geometrical information to train a predictive model for m6A site identification.