Generative AI-Powered Exploration of Protein Dynamics and Non-Deterministic Chemical Kinetics - PROJECT SUMMARY Overview. The PI’s lab focuses on uncovering the dynamic mechanisms underlying protein functions through advanced computational simulations and theoretical modeling, employing molecular mechanics (MM), quantum mechanics (QM), and their combined approaches (QM/MM). The group's research has concentrated on three major themes. Protein Allostery: Significant advances have been achieved in both methodological development and application studies of key allosteric proteins, including the Light-Oxygen-Voltage-sensing (LOV) domain and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2). Additionally, the group has created two user- friendly public computational servers to serve the broader research community. Evolution of Enzyme Catalysis: The PI’s lab has pioneered innovative theoretical frameworks that integrate machine learning techniques with catalytic reaction pathways derived from hybrid QM/MM calculations, enhancing our understanding of enzyme evolution and catalytic mechanisms. Fundamental Chemical Mechanism Studies: The PI has a longstanding interest in the mechanisms underlying chemical transitions, particularly focusing on their characterization as rare events. Through extensive computational investigations and innovative methodological advancements, the PI's laboratory has explored a broad spectrum of chemical processes in detail. Goals of Next Five Years. Over the next five years, our research will advance computational modeling across multiple cutting-edge domains, integrating artificial intelligence and advanced theoretical frameworks to address complex biochemical and photochemical phenomena. We aim to establish transformer-based conditional diffusion models to predict protein dynamics, creating an automated, scalable alternative to traditional molecular dynamics simulations. Concurrently, we will expand our generative AI approaches to dissect catalytic mechanisms of β-lactamase enzymes, systematically mapping the catalytic landscapes of serine-based and metallo-β-lactamases. By incorporating protein conformational dynamics into these catalytic models, we aim to elucidate enzyme regulation and evolution, enhancing understanding of antibiotic resistance mechanisms. Additionally, we will develop comprehensive stochastic differential equation (SDE) models using Itô calculus to characterize the stochastic molecular behaviors underlying mechanically activated photochemistry (MAPC). Integrating specialized models for chemiluminescence and mechanochemical activation into a unified MAPC framework, we aim to significantly enhance predictive simulation capabilities in this interdisciplinary area. Collectively, these efforts will produce robust theoretical and computational tools that deepen fundamental understanding and enable innovative applications across biophysics and biochemistry. Overall Vision. Understanding the dynamic behavior of proteins is central to decoding the mechanisms underpinning protein function, enzymatic catalysis, and structure-function relationships. It remains computationally prohibitive for simulating biophysical and biochemical processes at functionally relevant timescales, which often span seconds to minutes—far beyond the millisecond range achievable by current technologies. Chemical processes at the molecular level are inherently non-deterministic stochastic processes, governed by probabilistic events and random molecular fluctuations rather than fixed, reproducible trajectories. The PI has been developing a long-term vision to advance biomolecular modeling by shifting from traditional MD simulations and reaction pathway calculation for enzyme catalysis mechanism toward AI-driven, predictive models capable of generating protein dynamics, enzyme catalytic mechanisms, and stochastic biochemical behaviors. By combining generative AI, molecular simulations, and stochastic modeling, this research addresses critical limitations in scaling biomolecular dynamics studies. It promises t
