Deep Generative Markov State Models with Experimental Restraints
Abstract
A central challenge in molecular modeling is reconciling simulations with experimental observables, as force-field inaccuracies can distort equilibrium populations and long-timescale kinetics. While time-dependent restraints can improve simulated kinetics, time-dependent structural observables remain limited. Thus, most experimental data are time-averaged, informing equilibrium ensembles but not directly dynamics. Although ensemble refinement can improve thermodynamic accuracy, incorporating time-averaged observables into kinetic models remains difficult. Here, we introduce BICePs-reweighted Reversible DeepMSMs, combining Bayesian Inference of Conformational Populations (BICePs), variational learning of Markov processes, and maximum entropy (MaxEnt)/maximum caliber (MaxCal) principles to infer consistent thermodynamics and minimally perturbed kinetics. At its core is a reversible DeepMSM prior, obtained by applying an orthogonal transformation post-hoc to a pre-trained dynamics model (e.g., VAMPnet). This preserves the eigenspectrum exactly while yielding a valid transition matrix that is nonnegative, row-stochastic, and reversible. The reweighted model refines stationary populations, transition dynamics, and state-conditioned configurational landing densities. We validate the approach on a quadruple-well toy system and alanine dipeptide using backbone dihedral angles and J-coupling constants. In both systems, perturbing the observables induces predictable changes in the equilibrium ensemble and corresponding inferred kinetics. The resulting relaxation timescales and dynamical modes closely agree with analytical and MaxCal reference models. Furthermore, a diffusion-based generative model trained on MaxEnt landing densities produces physically realistic alanine dipeptide trajectories that reproduce thermodynamics and kinetics while satisfying the imposed experimental restraints.
Related
- Ab initio phasing based on topological restraints: automated determination of the space group and the number of molecules in the unit cell
- Describing the Fluxional Dynamics of Sub-nanoscale Clusters with Markov State Modeling
- Robust generative transition-state models for unseen chemistry
- From Silicon to Carbon: Experimental Validation of Generative AI-Derived CNS Drug Candidates — A Systematic Review (2021–2026)
- Photoreactivity of E-N,N-bis(2-hydroxyethyl)cinnamamide in Methanol and Solid State: A Complementary Experimental and Theoretical Investigation
BibTeX
@misc{raddi2026deep,
title = {{Deep Generative Markov State Models with Experimental Restraints}},
author = {Robert M. Raddi and Vincent A. Voelz},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.27236},
eprint = {2609.27236},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.27236},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - Deep Generative Markov State Models with Experimental Restraints
AU - Robert M. Raddi
AU - Vincent A. Voelz
PY - 2026
DA - 2026/09/23
PB - arXiv
DO - 10.48550/arxiv.2609.27236
UR - https://arxiv.org/abs/2609.27236
AB - A central challenge in molecular modeling is reconciling simulations with experimental observables, as force-field inaccuracies can distort equilibrium populations and long-timescale kinetics. While time-dependent restraints can improve simulated kinetics, time-dependent structural observables remain limited. Thus, most experimental data are time-averaged, informing equilibrium ensembles but not directly dynamics. Although ensemble refinement can improve thermodynamic accuracy, incorporating time-averaged observables into kinetic models remains difficult. Here, we introduce BICePs-reweighted Reversible DeepMSMs, combining Bayesian Inference of Conformational Populations (BICePs), variational learning of Markov processes, and maximum entropy (MaxEnt)/maximum caliber (MaxCal) principles to infer consistent thermodynamics and minimally perturbed kinetics. At its core is a reversible DeepMSM prior, obtained by applying an orthogonal transformation post-hoc to a pre-trained dynamics model (e.g., VAMPnet). This preserves the eigenspectrum exactly while yielding a valid transition matrix that is nonnegative, row-stochastic, and reversible. The reweighted model refines stationary populations, transition dynamics, and state-conditioned configurational landing densities. We validate the approach on a quadruple-well toy system and alanine dipeptide using backbone dihedral angles and J-coupling constants. In both systems, perturbing the observables induces predictable changes in the equilibrium ensemble and corresponding inferred kinetics. The resulting relaxation timescales and dynamical modes closely agree with analytical and MaxCal reference models. Furthermore, a diffusion-based generative model trained on MaxEnt landing densities produces physically realistic alanine dipeptide trajectories that reproduce thermodynamics and kinetics while satisfying the imposed experimental restraints.
ER -