AI-Guided Protein Design for Programmable Biology

Our research program focuses on AI-guided protein design to engineer biological systems with unprecedented precision, spanning molecular discovery, receptor reprogramming, and de novo cellular engineering. We approach biology as a programmable system, using machine learning and generative models to design synthetic proteins, rewire signaling, and build new cellular functions from the ground up. Our long-term vision is to bridge artificial intelligence with synthetic biology to generate predictive and functional models of life at the molecular, network, and cellular level & translate those into real-world applications.

 

ML-based Discovery: Towards a Fully Predicted Virtual Cell

We aim to develop machine learning models that decode and predict the structure and function of biological systems. This includes generative approaches to protein design based on natural language prompts and foundation models that can capture single protein distributions across cellular states. In the long term, we envision a fully in silico virtual cell that can predict cellular responses, and enable biological design. Key directions involve pathway response shaping, large-scale structure-function prediction, and training generative models on curated annotation datasets.
 

Target: Designing Binders against any target of interest

Our work on binder design focuses on creating synthetic proteins that recognize defined epitopes for therapeutic and imaging applications. We aim to design binders against any epitope with the goal of targeting rare or disease-specific cells and not only inhibit but actively shape signaling response. 

Reprogram: Engineering Signaling Pathways

We use AI-guided protein design to reprogram cellular signaling at the receptor level. We aim to design receptors with defined clustering behaviors and signal propagation logic and are particularly interested in modulating signaling thresholds and pathway crosstalk by reshaping protein-protein interfaces.