Onko-Gene

Critical literature watch in the molecular oncology of solid tumors

FREN

Bioinfo & AI

RGCNMDA: pathway-bridged relational graph learning with adaptive multi-view fusion for human miRNA-disease association prediction

Summary

This study introduces RGCNMDA, a novel multi-view learning framework designed to predict microRNA-disease associations, specifically addressing challenges posed by sparse data and "cold-start" entities. The model integrates global latent structures, local profiles, similarities, and biological pathway context through a relational graph. It employs a relational graph convolutional network for message passing and adaptive fusion of representations. RGCNMDA achieved robust performance on HMDD v4.0 and RNADisease v4.0 benchmarks, showing high AUCs, especially in cold-start scenarios. The research highlights the significant contribution of pathway context, particularly for less characterized diseases.

Critical appraisal

RGCNMDA represents a significant advancement in microRNA-disease association prediction by proposing a relational graph architecture with adaptive multi-view fusion. The approach is novel due to its rigorous integration of biological pathway context and its controlled handling of data leakage, a common issue in this type of modeling. The model was extensively benchmarked on established datasets (HMDD v4.0, RNADisease v4.0) and demonstrated superior performance compared to baseline methods, especially in "cold-start" scenarios (new diseases or new miRNAs), which is crucial for biomarker discovery. The code is available on GitHub, promoting reproducibility and community adoption. If deployed clinically, such a tool could prioritize miRNA-disease pairs for experimental studies, accelerating the discovery of diagnostic or prognostic biomarkers, and potentially guiding targeted therapies by identifying relevant dysregulated miRNAs for complex diseases.

The summary and the appraisal were written by a language model from the title and abstract of the original article. A curator decided to publish it. Methodology