Abstract

Single-molecule Förster resonance energy transfer (smFRET) is a powerful technique for probing the structural dynamics of biomacromolecules. However, conventional analyses based on hidden Markov models (HMMs) impose a discrete-state framework that partitions conformational space into a limited number of states and models dynamics as transitions between them. While effective for identifying stable states and their kinetics, such approaches struggle to characterize transition states and continuous structural changes. Here, we present a Bayesian statistical framework that models smFRET time traces using the Langevin equation, enabling the inference of continuous conformational dynamics without the need for artificial intermediate states. Parameter estimation is performed using the Expectation–Maximization (EM) algorithm. A central feature of this approach is its ability to estimate transition-path times (TPTs), which capture the duration of barrier-crossing events and offer insight into transition processes that are otherwise inaccessible with discrete-state models. We validate the method using synthetic smFRET data to assess its accuracy and temporal resolution, and we further demonstrate its applicability to experimental datasets. This work provides a principled and flexible approach for reconstructing free-energy landscapes and transition dynamics from smFRET data, overcoming key limitations of traditional HMM -based analyses.