Machine learning-enhanced computational biophysics for molecular interaction prediction
Integrating computational biophysics with machine learning, this solution offers novel tools for virtual screening of compounds by accurately predicting molecular interactions and binding affinities, outperforming existing methods.
Overview
This innovative solution combines computational biophysics with machine learning to revolutionize the prediction of molecular interactions. By accurately predicting binding affinities, it provides a powerful tool for virtual screening of compounds, enabling the prioritization of molecules with desired properties. This approach outperforms existing methods, offering significant insights into peptide-protein and compound-molecule interactions.
Technical specifications
Key Features:
- System Preparation: Uses OpenMM for system setup.
- Simulation: Conducts 10 short metadynamics molecular dynamics simulations via colvars in GROMACS.
- Feature Extraction: Derives eight features, such as the average area under the hydrogen bond curve over time.
- Machine Learning Model: Trains models on extracted features using deep neural networks, with strategies to avoid overfitting.
Future Enhancements:
- Modify workflow to predict binding affinities between chemical compounds and molecules.
- Optimize parameters and simulation methods for improved predictions.
- Experiment with machine learning models to enhance accuracy.
Technology readiness level
Currently at TRL 3, the solution is in the experimental proof of concept stage, with plans for further development and optimization leveraging extensive research experience in calculating free energies for molecular systems.
About University of Idaho
Founded as Idaho’s land‑grant, the University of Idaho is a comprehensive public research university with a statewide footprint anchored in Moscow. Industry engages through a Post Falls research park that co‑locates companies with faculty, multi‑campus centers in Boise, Coeur d’Alene and Idaho Falls, and an Extension network active in 42 counties. Proximity and formal agreements with Idaho National Laboratory provide a direct conduit to national‑lab expertise and shared facilities in Idaho Falls. Research is backed by competitive federal funding from agencies such as the National Science Foundation, U.S. Department of Agriculture and the Department of Energy. A dedicated technology transfer office supports IP strategy, licensing and startup formation.
Inquire about this partnering listing
Industry applications
delivered to your inbox.