By combining artificial intelligence with molecular dynamics simulations, researchers at the Department of Energy’s Oak Ridge National Laboratory have created a powerful new tool to uncover the molecular basis of plant-microbe communication. This breakthrough enables more accurate predictions of how plants and beneficial microbes interact at the most fundamental level, paving the way for engineering smarter, more resilient crops. Plants and microbes rely on chemical signals called ligands to form symbiotic relationships that enhance plant growth, improve nutrient uptake, and increase resistance to stress. A key group of these signals, known as lipo-chitooligosaccharides (LCOs), plays a crucial role in establishing these partnerships. However, predicting how plant proteins recognize and bind to these large, flexible LCO molecules has long been difficult. Traditional tools like AlphaFold, while revolutionary for protein structure prediction, are limited when it comes to large, dynamic ligands and typically produce static models that don’t reflect real-world molecular motion. To overcome these challenges, ORNL scientists developed a hybrid approach called MD/ML—integrating molecular dynamics simulations, which model the constant movement and flexibility of proteins, with machine learning models trained on extensive datasets of protein-ligand interactions. This method captures the dynamic nature of molecular binding, offering a far more realistic picture than static predictions. The simulations were run on two of the world’s most powerful supercomputers—Frontier and Summit—at the Oak Ridge Leadership Computing Facility, a national user facility supporting open science and innovation. The results showed that the MD/ML approach could accurately predict binding strengths and structural details that matched experimental data, even when starting protein models were incomplete or low-resolution. “This method allows us to predict how strongly plant receptors bind to microbial signals, which directly influences which genes are activated and how plants respond physiologically,” said Omar Demerdash, co-lead of the project. “This is essential for understanding how plants build relationships with microbes—and how we can engineer better ones.” Erica Prates, also a project co-lead, emphasized the practical impact: “Being able to quickly predict these molecular ‘match-ups’ means we can focus lab experiments on the most promising candidates, saving time and resources.” Dan Jacobson, a corresponding author and computational systems biologist at ORNL, highlighted the importance of flexibility: “Proteins aren’t rigid—they’re constantly moving. Most prediction tools ignore this. Our method accounts for that motion, giving us a much more accurate view of how binding actually happens.” The new workflow not only accelerates research into plant-microbe interactions but also has broader applications in biotechnology, including designing better biofuels, reducing fertilizer use, and even repurposing existing drugs for new medical treatments. By decoding the molecular language of plant-microbe communication, this work supports national goals in energy security, sustainable agriculture, and scientific leadership.