Embedded Files
Research interests
AI for designing protein and biomolecular interactions
Protein design for targeting chemical modifications in biology
Enzyme design and engineering
Protein and nucleic acid interactions
Biosensors and diagnostics
Novel therapeutic platforms enabled by protein design
Computationally designing proteins with interfaces capable of binding specific chemical groups remains as a significant challenge. To address this, we developed an AI-based pipeline that integrates a protein structure generation model (RFdiffusion All Atom) and an amino acid sequence design model (LigandMPNN). This enabled the successful design of novel proteins binding and sensing diverse small-molecules.
We are expanding our design scope and are developing AI-based methods to design proteins capable of detecting and modulating chemical modifications in peptides and nucleic acids. Our current focus includes designing proteins that recognize specific glycosylation patterns, chemically modified RNA and peptides, as well as optimizing enzymes. By designing modification and context-specific binders and regulators, we aim to enable precise analysis and control of the biological systems governed by chemical modifications.
Our research integrates computational and experimental approaches. We aim to address protein design challenges that lack available data for AI model training, such as designing protein and RNA complexes and multi-state enzymes. By establishing a closed loop of AI-based design, experimental testing, and iterative model improvement, we seek to uncover the underlying principles of systems that demand tailored data generation across different methodological approaches.
Page updated
Google Sites
Report abuse