We build the technology of the future
Using the latest advances in AI, it is now possible to predict protein structures directly from sequence. Our work takes this further, enabling the structure prediction of interacting proteins and the assembly of large multi-protein complexes using reinforcement learning.
But structure is only the beginning.
Many biological functions are driven by interactions — between proteins, nucleic acids, small molecules, and engineered peptides. Using advanced deep learning, we are building tools to illuminate and design these interactions. With RareFold, it is now possible to predict and design proteins and peptides that contain noncanonical amino acids, expanding the chemical space far beyond what nature offers.
These capabilities have already been used to create functional molecules: dual cyclic peptide agonists that activate GCGR and GLP1R, and a cyclic peptide inhibitor of HIV-1 membrane fusion, designed in a single step. Deep learning is also guiding discovery by identifying previously hidden pathogenic protein interfaces, revealing new opportunities for therapeutic intervention.
Our goal is a universal molecular design framework where any molecule can be predicted and created for any application, at the click of a button.
Group Members:
Patrick Bryant, Principal investigator
Dr. David (Qiuzhen) Li – Postdoc
Diandra Daumiller – PhD student