AI-driven screening platform for identifying SGLT1 and GLUT5 modulators
Computational platform combining flexible docking with AI/ML models to predict activators and inhibitors of the SGLT1 and GLUT5 transporters. Uses structural data and activity databases to screen dietary ingredients and chemical libraries, enabling rapid identification of candidates for experimental testing in metabolic and nutrition research.
Overview
This solution is a computational screening platform designed to identify small-molecule modulators of two important human sugar transporters: SGLT1 (sodium-glucose cotransporter 1) and GLUT5 (a fructose transporter). These transporters play key roles in glucose and fructose absorption, making them attractive targets for metabolic disease research, nutritional science, and drug discovery.
The platform combines flexible molecular docking with artificial intelligence and machine learning to predict which compounds will activate or inhibit these transporters. By screening large libraries—including dietary ingredients—against validated models, the platform can rapidly prioritize the most promising candidates for experimental testing, significantly reducing the time and cost of discovery.
Technical specifications
Core approach:
- Flexible docking techniques based on Internal Coordinate Mechanics that account for induced fit between the transporter and its ligands
- Integration of the Pocketome, a curated collection of conformational states and co-crystal structures for thousands of human drug targets, including SGLT1 and GLUT5
- Hybrid docking and AI/ML models built from curated activity data in PubChem and ChEMBL, with careful separation of activators from inhibitors
- High-performance computing deployment using CPU cores and GPUs at the San Diego Supercomputer Center
Key capabilities:
- Predicts binding poses for any query molecule against SGLT1 or GLUT5
- Estimates activity (activation or inhibition) for novel compounds
- Screens large chemical collections, including food ingredients, to identify top candidates for experimental validation
Validation track record:
The underlying methodology has been validated in multiple peer-reviewed publications, including work on allosteric inhibitors, G-protein-coupled receptor ligands, endocrine disruptors, and the Pocketome framework itself.
Technology readiness level
The platform is built on validated docking and AI/ML techniques that have been published and applied across multiple drug discovery projects. Current activities focus on updating datasets for SGLT1 and GLUT5, rebuilding hybrid models with refined separation of activators and inhibitors, and deploying them in a high-performance computing environment. Experimental partners at UC San Diego are available to test top predictions. The platform is ready for collaborative application to specific chemical libraries and dietary ingredient lists, with feedback loops in place to iteratively improve model accuracy.
About University of California, San Diego
UC San Diego is a comprehensive public research university in La Jolla, anchored by an integrated academic health system. On the east campus, the 23‑acre Science Research Park co‑locates corporate R&D with university programs and connects to the health sciences district, enabling daily interaction among scientists, clinicians, and companies. The Jacobs School’s Corporate Affiliates Program provides structured engagement and recruiting, while the Office of Innovation & Commercialization supports IP, licensing, and startup formation. The research enterprise is sustained by competitive federal funding, with recent awards exceeding $1.7B annually. Established commercialization pathways streamline sponsored research and help partners move from proof of concept to deployment.
delivered to your inbox.