Oak Ridge National Laboratory AI ‑ MD predicts plant microbrigand binding

Computational systems biologist and corresponding author, Dan Jacobson of Oak Ridge National Laboratory, has demonstrated an AI-type workflow that predicts the binding of large-scale, flexible fatty chitrigosaccharide ligands to integrate molecular dynamics simulations and machine learning to achieve predictions tailored to plant receptor protons. Called MD/ML, this method is trained on a large dataset of protein ligand complexes and runs at Oak Ridge Leadership Computing facilities at the country's fastest supercomputers, frontiers and summits, ranking binding strengths even starting with rough protein models. By accurately predicting which plant genes govern the optimal microbial partnership, this approach promises to accelerate the engineering of the microbiota that allows crops to grow faster, reduce fertilizer, and generate more biomass through conversion to fuel, chemicals and materials.

Molecular dynamics for AI plant microbial signaling at Oak Ridge National Laboratory pioneer AI laboratory.

Oak Ridge National Laboratory (ORNL) announced on September 15, 2025 that it had introduced a hybrid workflow with long time-scale molecular dynamics (MD) simulations (ML) simulations to predict how plant receptors recognize lipochytrigosaccharides (LCOs). ORNL Computational and Predictive Biology Group and Co-Leaders Developed by the ORNL Molecular Biophysics Group under the Erica Prates and Omar Demerdash. This approach addresses the limitations of static tools such as AlphaFold, which optimizes for small drug-like ligands and misses the dynamic variation that governs LCO binding.

Frontier and Summit Supercomputers Power Dynamic Protein Ligand Prediction

The MD/ML workflow was run at Frontier and Summit, the country's fastest supercomputer at the Oak Ridge Leadership Computing facility. Extensive sampling of receptor flexibility captured transient binding events that would otherwise be missed, and the resulting predictions were consistent with experimental binding assays, revealing details of previously unknown structures at the ligand receptor interface. This framework can also probe hormonal signaling and pathogen defense pathways.

Ranking receptor-ligand interactions provide a mechanical map of plant genes that are most likely to modulate beneficial microbial partnerships. Leaded by Pleitz and Demardash, the teams including Toms Rush, Udayakalli and Maneschash demonstrated that engineered microbial consortiums can increase biomass yields to promote plant growth, reduce dependence on synthetic fertilizers, and expand biofelds to expand biofelds, biochemicals, and unified materials to expand biofelds, biochemicals, and unified materials. Biotechnology sector.

“Proteins are not rigid, they are constantly shaking,” said Dan Jacobson, a biologist with computational systems, and said that by incorporating MD simulations, the system can sample all receptor motions that govern ligand recognition. Quantitative binding strength scores generated by the MD/ML workflow expose the residues and conformations important for laboratory measurements and LCO involvement, allowing breeders and genetic engineers to prioritize receptors with the most powerful binding profile for crop improvement and explore the reuse of existing pharmaceutical treatments.

The competitiveness of US biotechnology and the global food security benefits of advanced plant microbial modeling

Supported by the DOE office of the Plant World Interface of the Plant World in the Science Focus Area of ​​the Science Focus Area, the ORNL initiative demonstrates how AI-driven MD ligand binding can translate molecular insights into the practical benefits of food security and industrial biotechnology. With rapid screening and prioritization of plant and miclove interactions, American companies are competitive in the international market for sustainable agricultural solutions, providing a versatile platform for biological discovery and broader life science applications.