AI in Bio and Health

AI in Bio

Biological function emerges from molecular interactions, conformational changes, and collective behavior. At MAIL, we combine molecular dynamics, statistical learning, and protein language models to study these mechanisms and develop computational tools for molecular discovery. Our work links sequence-based prediction with structure and dynamics, helping identify promising hypotheses for further validation.

Recent work includes AgentD (2026), a language-model agent that coordinates literature retrieval, molecular generation, property prediction, refinement, and protein–ligand evaluation for early-stage drug discovery. This connects our molecular modeling expertise with workflows that make specialized computational tools easier to use.

Our interdisciplinary collaborations also extend to tissue engineering, including airway organoids with extracellular matrix incorporation (Biomaterials, 2026), which investigate the organization and polarity of engineered biological tissues.

Featured Research Areas

🧪 Biomolecule–Synthetic Material Interactions

We simulate DNA/material interfaces using MD and extract meaningful collective variables using statistical learning. This uncovers how recognition and binding occur at the molecular level.

🔬 Protein–Small Molecule Binding & Conformational Dynamics

High-dimensional time-series from MD is reduced to low-dimensional reaction coordinates, helping us map conformational transitions critical to protein function.

🧠 Sequence-Based Property Prediction Using Language Models

We treat protein/ligand sequences like language and use transformer-based models to predict properties such as binding affinity without relying on 3D structure.

🧩 Dynamics-Informed Residue-Level Insights

We analyze residue-level dynamics to classify conformational switching events as stable or unstable—advancing our understanding of function and disorder in proteins.

Google Sites

Report abuse