Our research is in the area of Theoretical Chemistry and Molecular Biophysics. The operation of biological molecules is a highly dynamic process that relies on numerous functional conformational changes. We are interested in elucidating the dynamics of these conformational changes by developing new methods based on statistical mechanics that can bridge the gap between experiments and atomistic molecular dynamics (MD) simulations.
We believe that tight integration of analytical theories with modern computer simulations and machine learning techniques can lead to breakthroughs in many aspects of theoretical chemistry. We also believe that establishing links between molecular mechanisms of bio-systems and their cellular and genome-wide behaviors is key to understanding many fundamental biological processes.
To fulfill our vision, our group is currently working on the following directions:
- Theoretical Advances in Biomolecular Dynamics
A central theme of our group’s research is the development of rigorous statistical-mechanical theories and machine learning (ML) tools to understand biomolecular dynamics. Many essential biological functions are governed by conformational changes that are difficult to capture with experiments or conventional molecular simulations.
We helped establish Markov State Models (MSMs) as a standard framework for modeling long-timescale biomolecular conformational changes, but also identified fundamental limitations of MSMs arising from their Markovian assumption. To address systems with strong memory effects, we pioneered non-Markovian modeling approaches based on the Generalized Master Equation (GME). In 2020, we introduced the first practical GME framework with explicit memory kernels, enabling more accurate modeling of slowly relaxing biomolecular systems.
Building on this foundation, we developed the Integrative GME (IGME), which provides a stable, analytical solution to the GME once memory has decayed (J. Chem. Phys., 2023). IGME overcomes long-standing numerical challenges and enables efficient modeling of complex biomolecular dynamics.
These theoretical advances naturally led us toward physics-driven ML. We have developed ML frameworks that embed statistical mechanics principles directly into ML architectures. Notably, our memory-kernel–based neural networks (MEMnets) integrate IGME theory with deep learning to identify slow collective variables underlying long-timescale conformational changes (Nat. Comput. Sci. 2025). In related work, we introduced TS-DAR, a hyperspherical latent-space model capable of simultaneously identifying multiple transition states across complex free-energy landscapes, a long-standing challenge in computational chemistry (Nat. Commun. 2025). This research direction reflects our broader vision to establish new chemical theories and integrate them with modern ML architectures.
Key publications:
Liu, B., Cao S.,, Boysen, J.G., Xue, M., X. Huang, X.*,”Memory Kernel Minimization Based Neural Networks for Discovering Slow Collective Variables of Biomolecular Dynamics“, Nature Computational Science, 5, 562–571 , (2025)
Liu, B., Boysen, J.G., Unarta, I.C., Du, X., Li, Y., Huang, X.* “Exploring Transition States of Protein Conformational Changes via Out-of-Distribution Detection in the Hyperspherical Latent Space“, Nature Communications, 16, 349, (2025)
Cao, S., Qiu, Y., Kalin, M.L., Huang, X.*, “Integrative Generalized Master Equation: A Method to Study Long-timescale Biomolecular Dynamics via the Integrals of Memory Kernels”, J. Chem. Phys.,159, 134106, (2023)
Cao, S., Montoya-Castillo, Andres., Wang, W., Markland, T.*, Huang, X.*, “On the Advantages of Exploiting Memory in Markov State Models for Biomolecular Dynamics”, J. Chem. Phys.,153, 014105, (2020)
Konovalov, K.A., Unarta, I.S., Cao, S., Goonetilleke1, E.C., Huang, X.*, “Markov state models to study the functional dynamics of proteins in the wake of machine learning”, J. Am. Chem. Soc. Au, 1(9), 1330-1341, (2021)
- Mechanistic Understanding of Biomolecular Functions
A major application of the theoretical and computational tools developed by my group is the elucidation of mechanisms in complex biomolecular machines, with a major focus on RNA polymerase. Using MSMs, non-Markovian GME models, and physics-informed ML, we have revealed how RNA polymerases coordinate catalysis, proofreading, and transcriptional regulation. Our work has elucidated how RNA Polymerase II performs both nucleotide addition and cleavage within a single active site, how antibiotics inhibit bacterial RNA polymerase by blocking loading-gate dynamics, and how oxidative DNA damage bypasses transcriptional fidelity control, leading to error-prone transcription.
Beyond RNA polymerases, we collaborate closely with experimental groups to study allosteric regulation in signaling enzymes, including protein phosphatase 2A (PP2A), O-GlcNAc, and O-GlcNAc transferase. These studies reveal how disease-associated mutations reshape long-range allosteric pathways, providing mechanistic insight into dysregulation that is difficult to access experimentally.
Key publications:
Unarta, I.S., Cao, S., Kubo, S., Wang, W., Cheung, P.P.H., Gao, X., Takada, S., Huang, X.* “Role of Bacterial RNA Polymerase Gate Opening Dynamics in DNA Loading and Antibiotics Inhibition Elucidated by quasi-Markov State Model”, Proc. Nat. Acad. Sci. U.S.A., 118(17), e2024324118, (2021)
Tse,C.K.M., Xu, J., Xu, L., Sheong, F.K., Wang, S., Chow, H.Y., Gao, X., Li, X., Cheung, P.P.H.*, Wang, D.*, Zhang, Y.*, Huang, X.*, “Intrinsic Cleavage of RNA Polymerase II Adopts a Nucleobase-independent Mechanism Assisted by Transcript Phosphate “, Nature Catalysis, 2, 228–235, (2019)
Da, L., Pardo, F., Xu, L., Silva, D., Zhang, L., Gao, X., Wang, D.*, Huang, X.*“Bridge Helix Bending Promotes RNA Polymerase II Backtracking Through a Critical and Conserved Threonine Residue”, Nature Communications, 7, 11244, (2016)
Silva, D., Weiss. D., Pardo-Avila, F., Da, L., Levitt, M., Wang, D., Huang, X., “Millisecond Dynamics of RNA Polymerase II Translocation at Atomic Resolution”, Proc. Nat. Acad. Sci. U.S.A., 111, 7665-7670, (2014)
- Targeted Protein Degradation
Our group has recently expanded into targeted protein degradation (TPD), an emerging therapeutic strategy that eliminates disease-causing proteins rather than inhibiting them. PROTAC-based TPD relies on stabilizing weak and transient protein–protein interactions between a target protein and an E3 ubiquitin ligase—interactions that are difficult to capture with conventional modeling approaches. Using our IGME-based non-Markovian framework, we identified metastable interaction interfaces between KRAS and E3 ligases, providing physically grounded templates for the rational design of selective degraders.
Key publications:
Qiu, Y., Wiewiora, R.P., Izaguirre, J.A., Xu, H., Sherman, W., Tang, W.*, Huang, X.* “Non-Markovian Dynamic Models Identify Non-Canonical KRAS-VHL Encounter Complex Conformations for Novel PROTAC Design“, JACS Au, 4 (10), 3857–3868, (2024)
- New Theories for Biomolecular Solvation
Accurately modeling how water and ions interact with biomolecules remains a fundamental challenge in computational biophysics, with major implications for folding, binding, aggregation, and catalysis. Our group develops statistical-mechanical solvation theories based on three-dimensional reference interaction site theory (3DRISM) to provide physically grounded and computationally efficient alternatives to conventional implicit-solvent models. We introduced the 3DRISM-IDC model, which accurately captures hydration around highly charged biomolecules, and developed EPISOL and EPIPY to make advanced solvation modeling accessible to the broader community.
Key publications:
Swanson, P.C., Cao, S., Huang, X.*, “A Python tutorial for 3DRISM solvation calculations of chemical and biological molecules”, J. Chem. Phys., 163, 171501 (2025)
Cao, S.; Kalin, M.L.; Huang, X.*, “EPISOL: A Software Package with Expanded Functions to Perform 3D-RISM Calculations for the Solvation of Chemical and Biological Molecules“, J. Comput. Chem., 44, 1536-1549, (2023)
Cao, S.; Qiu, Y.; Unarta, I. C.; Goonetilleke, E.C.; Huang, X.*, “The Ion-Dipole Correction of the 3DRISM Solvation Model to Accurately Compute Water Distributions Around Negatively Charged Biomolecules“, J. Phys. Chem. B., 126(43), 8632, (2022)
Cao, S., Sheong, F.K., Huang, X., “Reference Interaction Site Model with Hydrophobicity induced density Inhomogeneity: An Analytical Theory to Compute Solvation Properties of Large Hydrophobic Solutes in the Mixture of Polyatomic Solvent Molecules”, J. Chem. Phys., 143, 054110, (2015)