Teenager Jerry Xu Wins $90,000 for AI Model Translating Protein Shapes into Numbers

Jerry Xu, a 17-year-old student at Lexington High School, was awarded $90,000 in the Regeneron Science Talent Search for developing an artificial-intelligence model that translates three-dimensional protein structures into strings of numbers. His work, called “A Unified Protein Embedding Model With Local and Global Structural Sensitivity,” aims to speed up the comparison of proteins by preserving key structural features in a compact numeric form, addressing both speed and detail-loss issues in existing methods.

The model could enable researchers to quickly match unknown protein regions with well-characterized ones, offering clues about their biological roles. Beyond his bioinformatics research, Xu captains his school’s math team, participates in the science bowl, and runs a summer lecture series for younger students, while also assisting a local chatbot project that explains municipal budgets. He also volunteers as a teaching assistant for a computer-science academy, illustrating his broader commitment to STEM outreach.