Dynamicasome: a molecular dynamics-guided and AI-driven pathogenicity prediction catalogue for all genetic mutations

2025-09-23

AI Summary

In genomic medicine, the pathogenicity of numerous genetic variants remains undetermined, and existing AI-based prediction tools exhibit limited performance on functionally validated datasets. To address this, we propose a novel paradigm integrating molecular dynamics (MD) simulations with deep learning: for the first time, high-dimensional conformational dynamic features—such as residue flexibility and hydrogen-bond network perturbations—derived from MD trajectories are explicitly embedded into neural network training to systematically dissect structure–function mechanisms underlying *PMM2* variants. Our model significantly outperforms state-of-the-art tools (e.g., AlphaMissense, EVE) in discriminating known pathogenic mutations and delivers interpretable pathogenicity predictions for multiple variants of uncertain significance (VUS). By moving beyond purely sequence- or static-structure-based modeling, this work establishes a new methodology for precision variant interpretation grounded in dynamic structural biology.

Technology Category

• Machine Learning: Deep Neural Architectures and Foundation Models
• Data Mining & Knowledge Management: Other Foundations of Data Mining & Knowledge Management
• Humans and AI: Other Foundations of Human Computation & AI

Application Category

• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Advances in genomic medicine accelerate the identi cation of mutations in disease-associated genes, but the pathogenicity of many mutations remains unknown, hindering their use in diagnostics and clinical decision-making. Predictive AI models are generated to combat this issue, but current tools display low accuracy when tested against functionally validated datasets. We show that integrating detailed conformational data extracted from molecular dynamics simulations (MDS) into advanced AI-based models increases their predictive power. We carry out an exhaustive mutational analysis of the disease gene PMM2 and subject structural models of each variant to MDS. AI models trained on this dataset outperform existing tools when predicting the known pathogenicity of mutations. Our best performing model, a neuronal networks model, also predicts the pathogenicity of several PMM2 mutations currently considered of unknown signi cance. We believe this model helps alleviate the burden of unknown variants in genomic medicine.

Problem

• Predicting pathogenicity of genetic mutations remains challenging with low accuracy
• Current AI tools perform poorly against functionally validated mutation datasets
• Unknown mutation significance hinders diagnostics and clinical decision-making

Innovation

• Integrating molecular dynamics simulations with AI
• Training neural networks on conformational data
• Predicting pathogenicity of unknown genetic mutations