Explainable RNA-Small Molecule Binding Affinity Prediction Based on Multiview Enhancement Learning.

Area of Science:

• Computational chemistry
• Drug discovery
• Bioinformatics

Background:

• RNA is a potential drug target, necessitating methods to screen RNA-small molecule interactions.
• Accurate prediction of RNA-small molecule binding affinity is a significant challenge in drug development.

Purpose of the Study:

• To develop an explainable deep learning model for predicting RNA-small molecule binding affinity.
• To address the challenges in accurately forecasting binding affinity using physicochemical and topological properties.

Main Methods:

• Proposed an explainable multiview, multiscale deep learning network (EMMPTNet).
• EMMPTNet utilizes four modules for efficient feature extraction from multiple data views.
• A multilayer perceptron predicts binding affinity from extracted multiview, multiscale features.

Main Results:

• EMMPTNet achieved a mean absolute error (MAE) of 0.058 and a Pearson correlation coefficient (PCC) of 0.773.
• The model demonstrated superior performance compared to current state-of-the-art methods.
• Analysis of feature extraction and importance visualization confirmed model interpretability.

Conclusions:

• EMMPTNet offers an effective and interpretable approach for predicting RNA-small molecule binding affinity.
• The model's generalization ability was confirmed through validation on novel compounds.
• This work advances computational drug discovery by improving the prediction of critical molecular interactions.