Pith. sign in

REVIEW 3 major objections 5 minor 136 references

Latent Thermodynamic Flows: Unified Representation Learning and Generative Modeling of Temperature-Dependent Behaviors from Limited Data

T0 review · 3 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read The paper claims that representation learning and generative modeling should be trained as one, and that the resulting latent model turns data from two simulated temperatures into free energy surfaces at any temperature, including an RNA…

desk verdict A genuinely useful integration of SPIB and normalizing flows with strong benchmarks, but the temperature-extrapolation claim has a concrete qualitative miss on LJ7 and the SI loss derivation contains an algebraic error. read the letter →

arxiv 2507.03174 v1 pith:TRGOZMFC submitted 2025-07-03 cs.LG cond-mat.stat-mechphysics.bio-phphysics.chem-ph

Yunrui Qiu , Richard John , Lukas Herron , Pratyush Tiwary This is my paper
classification cs.LGcond-mat.stat-mechphysics.bio-phphysics.chem-ph MSC 68T07
keywords latentthermodynamicflowstemperature-steerablegenerativemodelstatepredictiveinformationbottlenecknormalizingcollectivevariablesfreeenergysurfaceRNAtetraloopmeltingmoleculardynamics
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Latent Thermodynamic Flows (LaTF) argues that representation learning and generative modeling in molecular science should be solved jointly rather than separately, and that doing so makes temperature nearly free. The paper's central claim is that a model trained on simulation data from just two temperatures can generate equilibrium distributions and free energy surfaces at any temperature in between and beyond, reproducing protein state populations and an RNA tetraloop's melting behavior without running simulations at those temperatures. LaTF unites the State Predictive Information Bottleneck, which learns two low-dimensional collective variables and labels metastable states, with a normalizing flow that maps the latent distribution onto an exponentially tilted Gaussian prior whose most-probable radius and variance grow linearly with temperature. A reader should care because, if this holds, thermodynamic interpolation across temperature no longer requires expensive sampling at every condition of interest.

What carries the argument

The load-bearing object is the exponentially tilted Gaussian prior $r_T(z,\tau) \propto \exp(\tau\|z\|)\,\exp(-\|z\|^2/2T)$, whose completed square reads $(\|z\| - T\tau)^2/(2T)$: the most-probable radius $T\tau$ and the variance $T$ both grow linearly with temperature, so every temperature is a different ring in the same two-dimensional latent plane, and the ring geometry is what lets priors at different temperatures overlap without one swallowing the other. This prior replaces the standard Gaussian (the $\tau = 0$ case) used by earlier generative samplers for temperature transfer. The rest of the machinery is the joint loss of Eq. 3, which couples a RealNVP normalizing flow $F_\theta$ to the State Predictive Information Bottleneck: the first term rewards predicting the future metastable state $y_{t+\Delta t}$ from flowed samples, while the second and third terms penalize mismatch between the flowed latent distribution and the tilted prior, with the flow's Jacobian making latent densities and hence free energies exactly computable.

What would settle it

Train LaTF at two temperatures, run a relatively short unbiased simulation at an intermediate temperature, encode that trajectory with the frozen encoder, and compare the encoded free-energy surface against the LaTF-generated surface at the same temperature using the paper's symmetric KL metric; the temperature-invariance premise predicts agreement as close as at the training temperatures. The paper already reports one instance of what failure looks like: in the Lennard-Jones cluster, LaTF only gradually melts the most-populated hexagonal state with rising temperature, whereas long MD simulations show it is sharply suppressed, a mismatch the authors attribute to the choice of tilt parameter.

Watch

Extended reading notes

Core claim

On its own terms, the paper establishes that a single two-dimensional latent space can carry both a kinetically meaningful representation and a generative model that is accurate at the trained temperatures and transferable across temperature. The encoder projects molecular configurations into the latent space, the decoder assigns them to long-lived metastable states, and a real-valued non-volume-preserving (RealNVP) normalizing flow relates the encoded distribution to a temperature-steerable tilted Gaussian prior $r_T(z,\tau)$. Because that prior is radially symmetric with its density concentrated on a ring of radius $T\tau$ and variance $T$, raising the temperature both inflates and shifts the high-density region of latent space, which is the mechanism by which the model predicts how free-energy basins broaden, shift, and repopulate at unseen temperatures. The benchmarks on Chignolin and the Lennard-Jones cluster show generated free-energy surfaces and state populations tracking long unbiased simulations across a range of temperatures, and the RNA GCAA tetraloop result is presented as the payoff: trained on 300 K and 400 K data alone, LaTF reconstructs the temperature-dependent structural ensemble and yields a melting temperature of about 88 °C, against 84 ± 10 °C from simulated tempering and about 71 °C from experiment.

Load-bearing premise

The prediction at unseen temperatures assumes that the same learned latent coordinates and the same map between latent space and prior space stay valid at every temperature, leaving the tilted Gaussian prior $r_T(z,\tau)$ to carry all temperature dependence; the authors state in the conclusion that there is no theoretical guarantee the learned representations remain unchanged with temperature.

Editorial extensions

If this is right

  • Two-temperature training replaces multi-temperature sampling: equilibrium distributions, free energy surfaces, and state populations at any temperature in the trained range are obtained by drawing from the tilted prior at that temperature and mapping back through the inverse flow.
  • One trained model simultaneously labels metastable states, interpolates transition pathways between basins, and back-maps generated latent samples to all-atom structures, so representation, generation, and interpretation no longer require separate pipelines.
  • The tilted Gaussian prior outperforms the standard temperature-scaled Gaussian on every benchmark, so the ring-shaped prior is essential to temperature transfer rather than a cosmetic choice.
  • Data efficiency persists under scarcity: training on 1 µs segments at two temperatures still captures the temperature-dependent free energy landscape of Chignolin, and training at two high temperatures still generates correct low-temperature ensembles.
  • For the RNA GCAA tetraloop, training at 300 K and 400 K reconstructs the continuous unfolding transition, giving a melting temperature of about 88 °C, close to the simulated tempering estimate of 84 ± 10 °C and somewhat above the experimental 71 °C.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the temperature-invariance premise holds, the same tilt mechanism should extend to other thermodynamic knobs such as pressure, chemical potential, or salt concentration by making each an axis of the prior; nothing in the architecture is specific to temperature.
  • Making the tilt parameter a learned function of temperature, rather than a constant chosen by validation error, is the natural next test: it would drop the paper's implicit assumption that one optimal tilt serves all temperatures and might correct known mismatches such as the Lennard-Jones cluster's over-populated hexagonal state at high temperature.
  • The framework inverts the compute tradeoff of molecular simulation: instead of converging simulations at every temperature, one invests once in a faithful latent representation and then re-samples it at nearly zero marginal cost, which pays off most on rough, glassy landscapes such as RNA where per-temperature sampling is the expensive step.
  • Because the prior's most-probable latent radius grows as $T\tau$, the model predicts that the radius of the encoded data cloud in latent space is close to linear in temperature; encoding unconverged intermediate-temperature data and checking that radius is a cheap, direct test of the premise.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The manuscript introduces Latent Thermodynamic Flows (LaTF), a framework that couples the State Predictive Information Bottleneck (SPIB) representation learning method with a normalizing flow (NF) and an exponentially tilted Gaussian prior. The central claim is that LaTF, trained on molecular dynamics data from only two temperatures, can generate equilibrium distributions and free energy surfaces at unseen temperatures by varying a temperature-steerable parameter in the tilted prior. The method is applied to a 2D potential, the Chignolin protein, a Lennard-Jones 7 cluster, and an RNA GCAA tetraloop, with benchmarks against long unbiased MD simulations, Markov state models, simulated tempering, and experiment. The paper emphasizes joint optimization of representation learning and generative modeling, improved metastable-state separation, physically realistic pathway interpolation, and data efficiency.

Significance. If the central claims hold, LaTF would be a valuable contribution: it provides an end-to-end pipeline that both learns interpretable collective variables and generates equilibrium ensembles at temperatures not directly simulated, potentially replacing expensive multi-temperature sampling for thermodynamic interpolation. The manuscript benefits from extensive empirical evaluation, including cross-validated KL divergences, GMRQ/MSM comparisons, out-of-sample temperature checks, and publicly released code. The temperature-extrapolation claim, however, rests on a strong structural assumption that is acknowledged by the authors but not validated, and one of the paper's own benchmark results provides a qualitative counterexample. The derivation of the unified loss also contains a technical error in dropping a θ-dependent term.

major comments (3)
  1. [SI §I.C, Eq. (12)-(13)] The derivation of the unified LaTF loss treats the encoder density p_θ(u|X) as a constant when moving from Eq. (11) to Eq. (12). Specifically, Eq. (11) contains the term −β log p_θ(u|X_n) inside the expectation over p_θ(u|X_n); this term is the differential entropy of the Gaussian encoder and depends on the network parameters θ, so it cannot be absorbed into the constant. Consequently, the claimed equivalence between L_LaTF in Eq. (3) of the main text and the original SPIB objective is not established. The actual objective being optimized omits an encoder-entropy regularization term (−β E[log p_θ(u|X)]), which may alter the balance between prediction accuracy and prior regularization. The authors should either correct the derivation, state the modified objective explicitly, or explain why the omitted term is benign in practice.
  2. [§II.C, Fig. 5(d) and Conclusion] The temperature-extrapolation mechanism assumes that the encoder and normalizing flow are temperature-invariant and that all temperature dependence of the equilibrium distribution is captured by the tilted Gaussian prior r_T(z,τ) in Eq. (5). The authors explicitly acknowledge in the Conclusion that there is no theoretical guarantee that learned representations remain unchanged with temperature. The LJ7 results provide a concrete falsification case: the main text states that the hexagonal metastable state 'is gradually diminished as temperature rises, whereas MD simulations show it should be sharply suppressed.' This is a qualitative failure of the predicted temperature trend in one of the benchmark systems, and the suggestion that refining τ would resolve it is not supported by any systematic τ-sensitivity analysis at unseen temperatures. This issue directly affects the central claim of predicting temperature-dependent behaviors at unseen temperatures and needs to be addressed, either by demonstrating when the assumption holds, by proposing a more general temperature-dependent transformation, or by restricting the claim to systems where the assumption is empirically verified.
  3. [§II.D, Fig. 6(e)] The validation of the RNA melting-curve prediction is presented largely through a single scalar quantity, the melting temperature (88°C for LaTF versus 84±10°C for simulated tempering and 71°C experimental). Since the predicted and reference curves are not compared quantitatively over the full temperature range (e.g., via overlapping confidence intervals or a mean-absolute-error metric), a correct Tm can coexist with substantial errors in the free energy surface at intermediate temperatures. In addition, the folded/unfolded classification depends on the nearest-neighbor backmapping in the latent space, whose accuracy at unseen temperatures is not separately quantified. Please provide a quantitative comparison of the full melting curves, including uncertainties propagated from the cross-validation, and clarify the effect of the backmapping approximation on the reported unfolded fractions.
minor comments (5)
  1. [Eq. (1) and Eq. (3)] The sign convention is confusing: Eq. (1) defines L_IB as a quantity to be maximized (with a leading minus), and Eq. (3) has a similar structure, but the text later refers to 'reconstruction loss' and 'generation error' as quantities to be minimized. Please clarify the optimization direction consistently.
  2. [§II.B, first sentence of 'Setting up LaTF'] Typo: 'the the IB latent dimension' should read 'the IB latent dimension'.
  3. [§II.C, 'Temperature-steerable tilted Gaussian prior'] The parameter τ is called both 'tilting factor' and 'temperature-steerable parameter' in different places. Since Eq. (5) also introduces the temperature T, it would help to reserve 'tilting factor' for τ and use a distinct name for the temperature input.
  4. [Methods, Eq. (8)] The normalization constant Z_{τ,T} is written with a factor τ√(2T); the corresponding expression in the SI (Eq. 26) appears to lack a factor of 2 in the numerator of the volume element. Please verify the consistency of these formulas.
  5. [§II.D, 'Exploring Temperature-Dependent RNA Free Energy Landscapes'] The sentence 'the discrepancy likely arises from factors such as force-field limitations, state definitions, or ionic conditions, etc.' is speculative; since the simulated-tempering reference uses the same force field, it would be more informative to quantify the systematic bias relative to experiment in a controlled way rather than listing possible causes.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the temperature dependence of the predicted ensembles is a stated forward-modeling assumption, and the paper's key benchmarks are external to the fitted parameters.

full rationale

The central claim is that LaTF, trained at two temperatures, can generate equilibrium distributions and free energy surfaces at unseen temperatures. The mechanism is explicit: the encoder and normalizing flow are trained once, and only the analytically defined tilted Gaussian prior r_T(z, tau) in Eq. 5 changes with temperature. This is a transparent modeling assumption, not a circular derivation: the predicted FES is, by construction, the prior pulled back through the fixed flow, but the prior is not fitted to the unseen-temperature data. The hyperparameter tau is selected using only KL divergence at the training temperatures (e.g., Fig. 4(b), Fig. 5(b), Fig. 6(b)), and the evaluations at unseen temperatures use full MD data only as a benchmark, not for parameter selection. For the RNA tetraloop, the predicted melting curve is compared against external experimental data and independent simulated tempering simulations using the same force field, which provides external falsifiability. The paper's own admission that 'there is no theoretical guarantee that the learned representations themselves should remain unchanged with temperature' is an honest statement of a scope limitation, not evidence of circularity. Self-citations to SPIB and to the authors' earlier Thermodynamic Maps work are used as building blocks and inspiration, but the unified loss is derived in the SI and the central benchmarks are external; no load-bearing reduction to a self-citation chain is present. The LJ7 hexagon-population discrepancy is an empirical failure of the temperature-invariance assumption, which further confirms that the predictions are not trivially enforced by construction. Under the stated rules, this is a non-finding: no circular step is identifiable.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The central claim rests on several fitted or assumed quantities. The tilting factor tau is fit to training-temperature validation data, the temperature-scaling law is assumed ad hoc, and the temperature invariance of the learned representation is explicitly acknowledged as unproven. These are the main costs the reader pays beyond the prior literature.

free parameters (5)
  • tilting factor tau = tau = 3 (2D potential), 2.5-3.5 (Chignolin), 2 (LJ7), 4.5 (RNA)
    Selected per system by minimizing symmetric KL between generated and validation-encoded IB distributions at training temperatures (Sec. II.B, II.C; SI Sec. V.C), so it is fitted to data and controls prior geometry.
  • Information bottleneck trade-off beta = 1e-4 to 5e-4 depending on system
    Chosen by hand in Table S2; controls the balance between future-state prediction and prior regularization.
  • lag time Delta t = 5 ns (Chignolin), 200 ns (RNA), 100 steps (LJ7), 1500 steps (model potential)
    Chosen per system and affects state definitions and the learned collective variables.
  • latent dimension d_z = 2
    Fixed to 2 for all systems following SPIB practice.
  • temperature normalization T = lowest training temperature set to 1; others proportional
    Ad hoc scaling of the steerable temperature parameter in r_T.
assumptions (4)
  • ad hoc to paper Tilted Gaussian prior ansatz: r_T(z, tau) captures all temperature dependence, with variance and most-probable radius linear in T
    Eq. 5 posits a one-parameter radial tilt as the full temperature response of the latent equilibrium distribution; no derivation from statistical mechanics is given.
  • domain assumption Latent representation and normalizing flow mapping are temperature-invariant
    The same encoder and flow are used for all temperatures; the conclusion explicitly states there is no theoretical guarantee for this.
  • domain assumption Nearest-neighbor matching in IB space is a valid backmapping to all-atom structures
    Generated latent samples are converted to structures using Euclidean nearest neighbors in the latent space, assuming latent distance correlates with structural similarity.
  • standard math RealNVP transformations are bijective with tractable Jacobians
    Standard normalizing flow change-of-variables result, used throughout the derivation.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Latent Thermodynamic Flows: Unified Representation Learning and Generative Modeling of Temperature-Dependent Behaviors from Limited Data." pith.science (2026). https://pith.science/paper/TRGOZMFC

@misc{pith2026250703174,
  author       = {Pith},
  title        = {Pith review of: Latent Thermodynamic Flows: Unified Representation Learning and Generative Modeling of Temperature-Dependent Behaviors from Limited Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TRGOZMFC}},
  note         = {Machine review of arXiv:2507.03174}
}
read the original abstract

Accurate characterization of the equilibrium distributions of complex molecular systems and their dependence on environmental factors such as temperature is essential for understanding thermodynamic properties and transition mechanisms. Projecting these distributions onto meaningful low-dimensional representations enables interpretability and downstream analysis. Recent advances in generative AI, particularly flow models such as Normalizing Flows (NFs), have shown promise in modeling such distributions, but their scope is limited without tailored representation learning. In this work, we introduce Latent Thermodynamic Flows (LaTF), an end-to-end framework that tightly integrates representation learning and generative modeling. LaTF unifies the State Predictive Information Bottleneck (SPIB) with NFs to simultaneously learn low-dimensional latent representations, referred to as Collective Variables (CVs), classify metastable states, and generate equilibrium distributions across temperatures beyond the training data. The two components of representation learning and generative modeling are optimized jointly, ensuring that the learned latent features capture the system's slow, important degrees of freedom while the generative model accurately reproduces the system's equilibrium behavior. We demonstrate LaTF's effectiveness across diverse systems, including a model potential, the Chignolin protein, and cluster of Lennard Jones particles, with thorough evaluations and benchmarking using multiple metrics and extensive simulations. Finally, we apply LaTF to a RNA tetraloop system, where despite using simulation data from only two temperatures, LaTF reconstructs the temperature-dependent structural ensemble and melting behavior, consistent with experimental and prior extensive computational results.

Figures

Figures reproduced from arXiv: 2507.03174 by the authors.

Figure 1
Figure 1. FIG. 1 [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2 [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3 [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: FIG. 4 [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5 [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6 [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

136 extracted references · 71 canonical work pages

  1. [1]

    State predictive information bottleneck

    Dedi Wang and Pratyush Tiwary. State predictive information bottleneck. The Journal of Chemical Physics , 154(13), 2021

    work page 2021

  2. [2]

    ) (19) =M(d 2, 1 2, 1 2τ 2) +τ √ 2Γ(d+1 2 ) Γ(d

    work page

  3. [3]

    Calculating protein–ligand residence times through state predictive information bottleneck based enhanced sampling

    Suemin Lee, Dedi Wang, Markus A Seeliger, and Pratyush Tiwary. Calculating protein–ligand residence times through state predictive information bottleneck based enhanced sampling. Journal of Chemical Theory and Computation , 20(14):6341–6349, 2024

    work page 2024

  4. [4]

    Alphafold2-rave: From sequence to boltzmann ranking

    Bodhi P Vani, Akashnathan Aranganathan, Dedi Wang, and Pratyush Tiwary. Alphafold2-rave: From sequence to boltzmann ranking. Journal of chemical theory and computation , 19(14):4351–4354, 2023

    work page 2023

  5. [5]

    The information bottleneck method

    Naftali Tishby, Fernando C Pereira, and William Bialek. The information bottleneck method. arXiv preprint physics/0004057, 2000

    arXiv 2000

  6. [6]

    Deep variational information bottleneck

    Alexander A Alemi, Ian Fischer, Joshua V Dillon, and Kevin Murphy. Deep variational information bottleneck. arXiv preprint arXiv:1612.00410, 2016

    arXiv 2016

  7. [7]

    Auto-encoding variational bayes, 2013

    Diederik P Kingma, Max Welling, et al. Auto-encoding variational bayes, 2013

    2013

  8. [8]

    f-VAEs: Improve VAEs with Conditional Flows

    Jianlin Su and Guang Wu. f-vaes: Improve vaes with conditional flows. arXiv preprint arXiv:1809.05861 , 2018

    work page Pith review arXiv 2018

Show all 136 references
  1. [9]

    Score-based generative modeling in latent space

    Arash Vahdat, Karsten Kreis, and Jan Kautz. Score-based generative modeling in latent space. Advances in neural information processing systems , 34:11287–11302, 2021

    2021

  2. [10]

    Flow matching in latent space

    Quan Dao, Hao Phung, Binh Nguyen, and Anh Tran. Flow matching in latent space. arXiv preprint arXiv:2307.08698 , 2023

    2023 arXiv

  3. [11]

    The tilted variational autoencoder: Improving out-of-distribution detec- tion

    Griffin Floto, Stefan Kremer, and Mihai Nica. The tilted variational autoencoder: Improving out-of-distribution detec- tion. In The Eleventh International Conference on Learning Representations , 2023

    2023

  4. [12]

    Boltzmann generators: Sampling equilibrium states of many-body systems with deep learning

    Frank No´ e, Simon Olsson, Jonas K¨ ohler, and Hao Wu. Boltzmann generators: Sampling equilibrium states of many-body systems with deep learning. Science, 365(6457):eaaw1147, 2019

    2019

  5. [13]

    Inferring phase transitions and critical exponents from limited observations with thermodynamic maps

    Lukas Herron, Kinjal Mondal, John S Schneekloth Jr, and Pratyush Tiwary. Inferring phase transitions and critical exponents from limited observations with thermodynamic maps. Proceedings of the National Academy of Sciences , 121(52):e2321971121, 2024

    2024

  6. [14]

    From data to noise to data for mixing physics across temperatures with generative artificial intelligence

    Yihang Wang, Lukas Herron, and Pratyush Tiwary. From data to noise to data for mixing physics across temperatures with generative artificial intelligence. Proceedings of the National Academy of Sciences , 119(32):e2203656119, 2022

    2022

  7. [15]

    Density estimation using real nvp

    Laurent Dinh, Jascha Sohl-Dickstein, and Samy Bengio. Density estimation using real nvp. arXiv preprint arXiv:1605.08803, 2016

    2016 arXiv

  8. [16]

    Masked autoregressive flow for density estimation

    George Papamakarios, Theo Pavlakou, and Iain Murray. Masked autoregressive flow for density estimation. Advances in neural information processing systems , 30, 2017

    2017

  9. [17]

    Stochastic normalizing flows.Advances in neural information processing systems , 33:5933–5944, 2020

    Hao Wu, Jonas K¨ ohler, and Frank No´ e. Stochastic normalizing flows.Advances in neural information processing systems , 33:5933–5944, 2020

    2020

  10. [18]

    Glow: Generative flow with invertible 1x1 convolutions

    Durk P Kingma and Prafulla Dhariwal. Glow: Generative flow with invertible 1x1 convolutions. Advances in neural information processing systems , 31, 2018

    2018

  11. [19]

    Neural spline flows

    Conor Durkan, Artur Bekasov, Iain Murray, and George Papamakarios. Neural spline flows. Advances in neural infor- mation processing systems , 32, 2019

    2019

  12. [20]

    Efficient molecular dynamics using geodesic integration and solvent–solute splitting

    Benedict Leimkuhler and Charles Matthews. Efficient molecular dynamics using geodesic integration and solvent–solute splitting. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences , 472(2189):20160138, 2016

    2016

  13. [21]

    10 residue folded peptide designed by segment statistics

    Shinya Honda, Kazuhiko Yamasaki, Yoshito Sawada, and Hisayuki Morii. 10 residue folded peptide designed by segment statistics. Structure, 12(8):1507–1518, 2004

    2004

  14. [22]

    Structure and dynamics of the tip3p, spc, and spc/e water models at 298 k

    Pekka Mark and Lennart Nilsson. Structure and dynamics of the tip3p, spc, and spc/e water models at 298 k. The Journal of Physical Chemistry A , 105(43):9954–9960, 2001

    2001

  15. [23]

    Development and testing of the opls all-atom force field on conformational energetics and properties of organic liquids

    William L Jorgensen, David S Maxwell, and Julian Tirado-Rives. Development and testing of the opls all-atom force field on conformational energetics and properties of organic liquids. Journal of the american chemical society , 118(45):11225– 11236, 1996

    1996

  16. [24]

    Improved peptide and protein torsional energetics with the opls-aa force field

    Michael J Robertson, Julian Tirado-Rives, and William L Jorgensen. Improved peptide and protein torsional energetics with the opls-aa force field. Journal of chemical theory and computation , 11(7):3499–3509, 2015

    2015

  17. [25]

    An evaluation of force field accuracy for the mini-protein chignolin using markov state models

    Tim Marshall, Robert Raddi, and Vincent Voelz. An evaluation of force field accuracy for the mini-protein chignolin using markov state models. Biophysical Journal, 122(3):420a, 2023

    2023

  18. [26]

    Particle mesh ewald: An n log (n) method for ewald sums in large systems

    Tom Darden, Darrin York, Lee Pedersen, et al. Particle mesh ewald: An n log (n) method for ewald sums in large systems. Journal of chemical physics , 98:10089–10089, 1993

    1993

  19. [27]

    Lincs: A linear constraint solver for molecular simulations

    Berk Hess, Henk Bekker, Herman JC Berendsen, and Johannes GEM Fraaije. Lincs: A linear constraint solver for molecular simulations. Journal of computational chemistry , 18(12):1463–1472, 1997

    1997

  20. [28]

    Molecular dynamics with coupling to an external bath

    Herman JC Berendsen, JPM van Postma, Wilfred F Van Gunsteren, ARHJ DiNola, and Jan R Haak. Molecular dynamics with coupling to an external bath. The Journal of chemical physics , 81(8):3684–3690, 1984. 25

    1984

  21. [29]

    Strain fluctuations and elastic constants

    M Parrinello and A Rahman. Strain fluctuations and elastic constants. The Journal of Chemical Physics , 76(5):2662– 2666, 1982

    1982

  22. [30]

    Gromacs: High performance molecular simulations through multi-level parallelism from laptops to supercomputers

    Mark James Abraham, Teemu Murtola, Roland Schulz, Szil´ ard P´ all, Jeremy C Smith, Berk Hess, and Erik Lindahl. Gromacs: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX, 1:19–25, 2015

    2015

  23. [31]

    T. Löhr, K. Kohlhoff, G. T. Heller, C. Camilloni, and M. Vendruscolo, A kinetic ensemble of the alzheimer’s aβ peptide, Nature Computational Science1, 71 (2021)

    2021

  24. [32]

    S. M. Marques, P. Kouba, A. Legrand, J. Sedlar, L. Disson, J. Planas-Iglesias, Z. Sanusi, A. Kunka, J. Damborsky, T. Pajdla, et al., Covampnet: com- parative markov state analysis for studying effects of drug candidates on disordered biomolecules, JACS Au 4, 2228 (2024)

    2024

  25. [33]

    Y. Qiu, I. Jang, X. Huang, and A. Yethiraj, Unsuper- vised learning of structural relaxation in supercooled liquids from short-term fluctuations, Proceedings of the National Academy of Sciences122, e2427246122 (2025)

    2025

  26. [34]

    W. Chen, H. Sidky, and A. L. Ferguson, Nonlinear discovery of slow molecular modes using state-free re- versible vampnets, The Journal of chemical physics150 (2019)

    2019

  27. [35]

    D. Wang, Y. Qiu, E. R. Beyerle, X. Huang, and P. Tiwary, Information bottleneck approach for markov 14 model construction, Journal of chemical theory and computation 20, 5352 (2024)

    2024

  28. [36]

    Pengmei, C

    Z. Pengmei, C. Lorpaiboon, S. C. Guo, J. Weare, and A. R. Dinner, Using pretrained graph neural networks with token mixers as geometric featurizers for confor- mational dynamics, The Journal of Chemical Physics 162 (2025)

    2025

  29. [37]

    Wu and F

    H. Wu and F. Noé, Reaction coordinate flows for model reduction of molecular kinetics, The Journal of Chemi- cal Physics 160 (2024)

    2024

  30. [38]

    D. Teng, V. J. Meraz, A. Aranganathan, X. Gu, and P. Tiwary, AlphaFold2-RAVE: Protein Ensemble Generation with Physics-Based Sampling, ChemRxiv 10.26434/chemrxiv-2025-q3mwr-v2 (2025), preprint. This content has not been peer-reviewed

    2025 doi

  31. [39]

    X. Gu, A. Aranganathan, and P. Tiwary, Empowering alphafold2 for protein conformation selective drug dis- covery with alphafold2-rave, Elife13, RP99702 (2024)

    2024

  32. [40]

    F. Noé, S. Olsson, J. Köhler, and H. Wu, Boltzmann generators: Sampling equilibrium states of many-body systems with deep learning, Science 365, eaaw1147 (2019)

    2019

  33. [41]

    Köhler, L

    J. Köhler, L. Klein, and F. Noé, Equivariant flows: ex- act likelihood generative learning for symmetric densi- ties, International conference on machine learning , 5361 (2020)

    2020

  34. [42]

    Dibak, L

    M. Dibak, L. Klein, A. Krämer, and F. Noé, Tempera- ture steerable flows and boltzmann generators, Physical Review Research 4, L042005 (2022)

    2022

  35. [43]

    Klein and F

    L. Klein and F. Noé, Transferable boltzmann genera- tors, arXiv preprint arXiv:2406.14426 (2024)

    2024 arXiv

  36. [44]

    Zheng, J

    S. Zheng, J. He, C. Liu, Y. Shi, Z. Lu, W. Feng, F. Ju, J. Wang, J. Zhu, Y. Min,et al., Predicting equilibrium distributions for molecular systems with deep learning, Nature Machine Intelligence6, 558 (2024)

    2024

  37. [45]

    Moqvist, W

    S. Moqvist, W. Chen, M. Schreiner, F. Nüske, and S. Olsson, Thermodynamic interpolation: A generative approach to molecular thermodynamics and kinetics, Journal of Chemical Theory and Computation21, 2535 (2025)

    2025

  38. [46]

    Y. Wang, L. Herron, and P. Tiwary, From data to noise to data for mixing physics across temperatures with generative artificial intelligence, Proceedings of the Na- tional Academy of Sciences119, e2203656119 (2022)

    2022

  39. [47]

    Herron, K

    L. Herron, K. Mondal, J. S. Schneekloth Jr, and P. Ti- wary, Inferring phase transitions and critical exponents from limited observations with thermodynamic maps, Proceedings of the National Academy of Sciences121, e2321971121 (2024)

    2024

  40. [48]

    S. Lee, R. Wang, L. Herron, and P. Tiwary, Expo- nentially tilted thermodynamic maps (exptm): Predict- ing phase transitions across temperature, pressure, and chemical potential, arXiv preprint arXiv:2503.15080 (2025)

    2025 arXiv

  41. [49]

    E. R. Beyerle and P. Tiwary, Inferring the isotropic- nematic phase transition with generative machine learn- ing, arXiv preprint arXiv:2410.21034 (2024)

    2024 arXiv

  42. [50]

    A. B. Tsybakov and A. B. Tsybakov, Nonparametric estimators, Introduction to Nonparametric Estimation , 1 (2009)

    2009

  43. [51]

    Biroli, T

    G. Biroli, T. Bonnaire, V. De Bortoli, and M. Mézard, Dynamical regimes of diffusion models, Nature Commu- nications 15, 9957 (2024)

    2024

  44. [52]

    R. John, L. Herron, and P. Tiwary, A comparison of probabilistic generative frameworks for molecular simu- lations, The Journal of Chemical Physics162 (2025)

    2025

  45. [53]

    Vahdat, K

    A. Vahdat, K. Kreis, and J. Kautz, Score-based genera- tive modeling in latent space, Advances in neural infor- mation processing systems34, 11287 (2021)

    2021

  46. [54]

    Rombach, A

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, High-resolution image synthesis with latent diffusion models, Proceedings of the IEEE/CVF confer- ence on computer vision and pattern recognition , 10684 (2022)

    2022

  47. [57]

    Papamakarios, E

    G. Papamakarios, E. Nalisnick, D. J. Rezende, S. Mo- hamed, and B. Lakshminarayanan, Normalizing flows for probabilistic modeling and inference, Journal of Ma- chine Learning Research22, 1 (2021)

    2021

  48. [59]

    Wirnsberger, A

    P. Wirnsberger, A. J. Ballard, G. Papamakar- ios, S. Abercrombie, S. Racanière, A. Pritzel, D. Jimenez Rezende, and C. Blundell, Targeted free en- ergy estimation via learned mappings, The Journal of Chemical Physics 153 (2020)

    2020

  49. [60]

    Rizzi, P

    A. Rizzi, P. Carloni, and M. Parrinello, Targeted free energy perturbation revisited: Accurate free energies from mapped reference potentials, The journal of phys- ical chemistry letters12, 9449 (2021)

    2021

  50. [61]

    Olehnovics, Y

    E. Olehnovics, Y. M. Liu, N. Mehio, A. Y. Sheikh, M. R. Shirts, and M. Salvalaglio, Assessing the accuracy and efficiency of free energy differences obtained from reweighted flow-based probabilistic generative models, Journal of Chemical Theory and Computation20, 5913 (2024)

    2024

  51. [62]

    van Leeuwen, A

    S. van Leeuwen, A. P. d. A. Ortíz, and M. Dijkstra, A boltzmann generator for the isobaric-isothermal ensem- ble, arXiv preprint arXiv:2305.08483 (2023)

    2023 arXiv

  52. [63]

    V. J. Meraz, Z. Zou, and P. Tiwary, Simulating crys- tallization in a colloidal system using state predictive information bottleneck based enhanced sampling, The Journal of Physical Chemistry B128, 8207 (2024)

    2024

  53. [64]

    Floto, S

    G. Floto, S. Kremer, and M. Nica, The tilted variational autoencoder: Improving out-of-distribution detection, The Eleventh International Conference on Learning Representations (2023)

    2023

  54. [65]

    Prinz, H

    J.-H. Prinz, H. Wu, M. Sarich, B. Keller, M. Senne, M.Held, J.D.Chodera, C.Schütte,andF.Noé,Markov models of molecular kinetics: Generation and valida- tion, The Journal of chemical physics134 (2011)

    2011

  55. [66]

    R. T. McGibbon and V. S. Pande, Variational cross- validation of slow dynamical modes in molecular kinet- ics, The Journal of chemical physics142 (2015)

    2015

  56. [67]

    Shoemake, Animating rotation with quaternion curves, Proceedings of the 12th annual conference on Computer graphics and interactive techniques , 245 (1985)

    K. Shoemake, Animating rotation with quaternion curves, Proceedings of the 12th annual conference on Computer graphics and interactive techniques , 245 (1985)

    1985

  57. [68]

    Honda, K

    S. Honda, K. Yamasaki, Y. Sawada, and H. Morii, 10 residue folded peptide designed by segment statistics, Structure 12, 1507 (2004). 15

    2004

  58. [69]

    Vanden-Eijnden, Transition path theory, Computer Simulations in Condensed Matter Systems: From Ma- terials to Chemical Biology Volume 1 , 453 (2006)

    E. Vanden-Eijnden, Transition path theory, Computer Simulations in Condensed Matter Systems: From Ma- terials to Chemical Biology Volume 1 , 453 (2006)

    2006

  59. [70]

    F. Noé, C. Schütte, E. Vanden-Eijnden, L. Reich, and T. R. Weikl, Constructing the equilibrium ensemble of folding pathways from short off-equilibrium simulations, Proceedings of the National Academy of Sciences106, 19011 (2009)

    2009

  60. [71]

    Y. Qiu, M. S. O’Connor, M. Xue, B. Liu, and X. Huang, An efficient path classification algorithm based on varia- tional autoencoder to identify metastable path channels for complex conformational changes, Journal of chemi- cal theory and computation19, 4728 (2023)

    2023

  61. [72]

    Trizio, P

    E. Trizio, P. Kang, and M. Parrinello, Everything every- where all at once: a probability-based enhanced sam- pling approach to rare events, Nature Computational Science , 1 (2025)

    2025

  62. [73]

    P. Kang, E. Trizio, and M. Parrinello, Computing the committor with the committor to study the transition state ensemble, Nature Computational Science 4, 451 (2024)

    2024

  63. [74]

    Metzner, F

    P. Metzner, F. Noé, and C. Schütte, Estimating the sampling error: Distribution of transition matrices and functions of transition matrices for given trajectory data, Physical Review E—Statistical, Nonlinear, and Soft Matter Physics80, 021106 (2009)

    2009

  64. [75]

    Janson, A

    G. Janson, A. Jussupow, and M. Feig, Deep generative modeling of temperature-dependent structural ensem- bles of proteins, bioRxiv , 2025 (2025)

    2025

  65. [76]

    G. A. Tribello, M. Ceriotti, and M. Parrinello, A self- learning algorithm for biased molecular dynamics, Pro- ceedings of the National Academy of Sciences 107, 17509 (2010)

    2010

  66. [77]

    Schwerdtfeger and D

    P. Schwerdtfeger and D. J. Wales, 100 years of the lennard-jones potential, Journal of Chemical Theory and Computation 20, 3379 (2024)

    2024

  67. [78]

    J. P. Sheehy, A. R. Davis, and B. M. Znosko, Ther- modynamic characterization of naturally occurring rna tetraloops, Rna 16, 417 (2010)

    2010

  68. [79]

    D. Tan, S. Piana, R. M. Dirks, and D. E. Shaw, Rna force field with accuracy comparable to state-of-the- art protein force fields, Proceedings of the National Academy of Sciences115, E1346 (2018)

    2018

  69. [80]

    Frauenfelder, S

    H. Frauenfelder, S. G. Sligar, and P. G. Wolynes, The energy landscapes and motions of proteins, Science254, 1598 (1991)

    1991

  70. [81]

    J. N. Onuchic, Z. Luthey-Schulten, and P. G. Wolynes, Theory of protein folding: the energy landscape per- spective, Annual review of physical chemistry48, 545 (1997)

    1997

  71. [82]

    Chen and K

    S.-J. Chen and K. A. Dill, Rna folding energy land- scapes, Proceedings of the National Academy of Sci- ences 97, 646 (2000)

    2000

  72. [83]

    Kuhrova, R

    P. Kuhrova, R. B. Best, S. Bottaro, G. Bussi, J. Sponer, M. Otyepka, and P. Banas, Computer folding of rna tetraloops: identification of key force field deficiencies, Journal of chemical theory and computation12, 4534 (2016)

    2016

  73. [84]

    Pagnani, G

    A. Pagnani, G. Parisi, and F. Ricci-Tersenghi, Glassy transition in a disordered model for the rna secondary structure, Physical review letters84, 2026 (2000)

    2000

  74. [85]

    Chakraborty, R

    D. Chakraborty, R. Collepardo-Guevara, and D. J. Wales, Energy landscapes, folding mechanisms, and ki- netics of rna tetraloop hairpins, Journal of the American Chemical Society 136, 18052 (2014)

    2014

  75. [86]

    G. H. Zerze, P. M. Piaggi, and P. G. Debenedetti, A computational study of rna tetraloop thermodynam- ics, including misfolded states, The Journal of Physical Chemistry B 125, 13685 (2021)

    2021

  76. [87]

    Rahimi, P

    K. Rahimi, P. M. Piaggi, and G. H. Zerze, Comparison of on-the-fly probability enhanced sampling and paral- lel tempering combined with metadynamics for atom- istic simulations of rna tetraloop folding, The Journal of Physical Chemistry B127, 4722 (2023)

    2023

  77. [88]

    Boccalini, Y

    M. Boccalini, Y. Berezovska, G. Bussi, M. Paloni, and A. Barducci, Exploring rna destabilization mechanisms in biomolecular condensates through atomistic simula- tions, Proceedings of the National Academy of Sciences 122, e2425261122 (2025)

    2025

  78. [89]

    Bottaro, A

    S. Bottaro, A. Gil-Ley, and G. Bussi, Rna folding path- ways in stop motion, Nucleic acids research 44, 5883 (2016)

    2016

  79. [90]

    F. M. Jucker, H. A. Heus, P. F. Yip, E. H. Moors, and A. Pardi, A network of heterogeneous hydrogen bonds in gnra tetraloops, Journal of molecular biology 264, 968 (1996)

    1996

  80. [91]

    Vangaveti, S

    S. Vangaveti, S. V. Ranganathan, and A. A. Chen, Ad- vances in rna molecular dynamics: a simulator’s guide to rna force fields, Wiley Interdisciplinary Reviews: RNA 8, e1396 (2017)

    2017

  81. [92]

    Sponer, G

    J. Sponer, G. Bussi, M. Krepl, P. Banáš, S. Bottaro, R. A. Cunha, A. Gil-Ley, G. Pinamonti, S. Poblete, P. Jurecka,et al., Rna structural dynamics as captured by molecular simulations: a comprehensive overview, Chemical reviews 118, 4177 (2018)

    2018

  82. [93]

    Y. I. Yang and M. Parrinello, Refining collective coordi- nates and improving free energy representation in vari- ational enhanced sampling, Journal of chemical theory and computation 14, 2889 (2018)

    2018

  83. [94]

    Frassek, A

    M. Frassek, A. Arjun, and P. Bolhuis, An extended au- toencoder model for reaction coordinate discovery in rare event molecular dynamics datasets, The Journal of Chemical Physics155 (2021)

    2021

  84. [95]

    Lemke and C

    T. Lemke and C. Peter, Encodermap: Dimensional- ity reduction and generation of molecule conformations, Journal of chemical theory and computation15, 1209 (2019)

    2019

  85. [96]

    Lipman, R

    Y. Lipman, R. T. Chen, H. Ben-Hamu, M. Nickel, and M. Le, Flow matching for generative modeling, arXiv preprint arXiv:2210.02747 (2022)

    2022 arXiv

  86. [97]

    J. Ho, A. Jain, and P. Abbeel, Denoising diffusion prob- abilistic models, Advances in neural information pro- cessing systems 33, 6840 (2020)

    2020

  87. [98]

    J.Song, C.Meng,andS.Ermon,Denoisingdiffusionim- plicit models, arXiv preprint arXiv:2010.02502 (2020)

    2020 arXiv

  88. [99]

    Metzner, C

    P. Metzner, C. Schütte, and E. Vanden-Eijnden, Illus- trationoftransitionpaththeoryonacollectionofsimple examples, The Journal of chemical physics125 (2006)

    2006

  89. [100]

    W. L. Jorgensen, D. S. Maxwell, and J. Tirado-Rives, Development and testing of the opls all-atom force field on conformational energetics and properties of organic liquids, Journal of the american chemical society118, 11225 (1996)

    1996

  90. [101]

    M. J. Robertson, J. Tirado-Rives, and W. L. Jorgensen, Improved peptide and protein torsional energetics with the opls-aa force field, Journal of chemical theory and computation 11, 3499 (2015). 16

    2015

  91. [102]

    Mark and L

    P. Mark and L. Nilsson, Structure and dynamics of the tip3p, spc, and spc/e water models at 298 k, The Jour- nal of Physical Chemistry A105, 9954 (2001)

    2001

  92. [103]

    I. L. Hofacker, W. Fontana, P. F. Stadler, L. S. Bonho- effer, M. Tacker, P. Schuster,et al., Fast folding and comparison of rna secondary structures, Monatshefte fur chemie 125, 167 (1994)

    1994

  93. [104]

    Lorenz, S

    R. Lorenz, S. H. Bernhart, C. Höner zu Siederdissen, H. Tafer, C. Flamm, P. F. Stadler, and I. L. Hofacker, Viennarna package 2.0, Algorithms for molecular biol- ogy 6, 1 (2011)

    2011

  94. [105]

    Das and D

    R. Das and D. Baker, Macromolecular modeling with rosetta, Annu. Rev. Biochem.77, 363 (2008)

    2008

  95. [106]

    R. Das, J. Karanicolas, and D. Baker, Atomic accuracy in predicting and designing noncanonical rna structure, Nature methods 7, 291 (2010)

    2010

  96. [107]

    A. M. Watkins, R. Rangan, and R. Das, Farfar2: im- proved de novo rosetta prediction of complex global rna folds, Structure 28, 963 (2020)

    2020

  97. [108]

    J. L. Abascal and C. Vega, A general purpose model for the condensed phases of water: Tip4p/2005, The Journal of chemical physics123 (2005)

    2005

  98. [109]

    NaritomiandS

    Y. NaritomiandS. Fuchigami,Slow dynamicsin protein fluctuations revealed by time-structure based indepen- dent component analysis: the case of domain motions, The Journal of chemical physics134 (2011)

    2011

  99. [110]

    Noé and C

    F. Noé and C. Clementi, Kinetic distance and kinetic maps from molecular dynamics simulation, Journal of chemical theory and computation11, 5002 (2015)

    2015

  100. [111]

    S.Bottaro, F.DiPalma,andG.Bussi,Theroleofnucle- obase interactions in rna structure and dynamics, Nu- cleic acids research42, 13306 (2014). Supplementary Information: Latent Thermodynamic Flows: Unified Representation Learning and Generative Modeling of Temperature-Dependent B...

    2014

  101. [112]

    When training generative models on data collected at different temperatures, the prior distribution should be designed to reflect temperature dependence

    M(d + 1 2 , 3 2, 1 2τ 2) (20) where M(a,b,z ) = ∑∞ n=0 a(n) b(n) zn n! denotes the Kummer confluent hypergeometric function, while a(n) = a(a + 1)··· (a +n− 1) represents the rising factorial. When training generative models on data collected at different temperatures, the pri...

  102. [113]

    (26) 7 III

    M(d + 1 2 , 3 2,T 2τ 2). (26) 7 III. NORMALIZING FLOWS AND REAL NVP TRANSFORMA TIONS Probabilistic generative models seek to approximate a target probability distribution from a finite number of samples and quickly generate new samples obeying this estimated density. Normalizi...

  103. [114]

    Information bottleneck approach for markov model construction

    Dedi Wang, Yunrui Qiu, Eric R Beyerle, Xuhui Huang, and Pratyush Tiwary. Information bottleneck approach for markov model construction. Journal of chemical theory and computation , 20(12):5352–5367, 2024

    2024

  104. [115]

    A self-learning algorithm for biased molecular dynamics

    Gareth A Tribello, Michele Ceriotti, and Michele Parrinello. A self-learning algorithm for biased molecular dynamics. Proceedings of the National Academy of Sciences , 107(41):17509–17514, 2010

    2010

  105. [116]

    100 years of the lennard-jones potential

    Peter Schwerdtfeger and David J Wales. 100 years of the lennard-jones potential. Journal of Chemical Theory and Computation, 20(9):3379–3405, 2024

    2024

  106. [117]

    A graph neural network-state predictive information bottleneck (gnn-spib) approach for learning molecular thermodynamics and kinetics

    Ziyue Zou, Dedi Wang, and Pratyush Tiwary. A graph neural network-state predictive information bottleneck (gnn-spib) approach for learning molecular thermodynamics and kinetics. Digital Discovery, 4(1):211–221, 2025

    2025

  107. [118]

    Plumed 2: New feathers for an old bird

    Gareth A Tribello, Massimiliano Bonomi, Davide Branduardi, Carlo Camilloni, and Giovanni Bussi. Plumed 2: New feathers for an old bird. Computer physics communications , 185(2):604–613, 2014

    2014

  108. [119]

    Fast folding and comparison of rna secondary structures

    Ivo L Hofacker, Walter Fontana, Peter F Stadler, L Sebastian Bonhoeffer, Manfred Tacker, Peter Schuster, et al. Fast folding and comparison of rna secondary structures. Monatshefte fur chemie , 125:167–167, 1994

    1994

  109. [120]

    Viennarna package 2.0

    Ronny Lorenz, Stephan H Bernhart, Christian H¨ oner zu Siederdissen, Hakim Tafer, Christoph Flamm, Peter F Stadler, and Ivo L Hofacker. Viennarna package 2.0. Algorithms for molecular biology , 6:1–14, 2011

    2011

  110. [121]

    Macromolecular modeling with rosetta

    Rhiju Das and David Baker. Macromolecular modeling with rosetta. Annu. Rev. Biochem. , 77(1):363–382, 2008

    2008

  111. [122]

    Atomic accuracy in predicting and designing noncanonical rna structure

    Rhiju Das, John Karanicolas, and David Baker. Atomic accuracy in predicting and designing noncanonical rna structure. Nature methods, 7(4):291–294, 2010

    2010

  112. [123]

    Farfar2: improved de novo rosetta prediction of complex global rna folds

    Andrew Martin Watkins, Ramya Rangan, and Rhiju Das. Farfar2: improved de novo rosetta prediction of complex global rna folds. Structure, 28(8):963–976, 2020

    2020

  113. [124]

    The role of nucleobase interactions in rna structure and dynamics

    Sandro Bottaro, Francesco Di Palma, and Giovanni Bussi. The role of nucleobase interactions in rna structure and dynamics. Nucleic acids research , 42(21):13306–13314, 2014

    2014

  114. [125]

    Rna force field with accuracy comparable to state-of- the-art protein force fields

    Dazhi Tan, Stefano Piana, Robert M Dirks, and David E Shaw. Rna force field with accuracy comparable to state-of- the-art protein force fields. Proceedings of the National Academy of Sciences , 115(7):E1346–E1355, 2018

    2018

  115. [126]

    ff14sb: improving the accuracy of protein side chain and backbone parameters from ff99sb

    James A Maier, Carmenza Martinez, Koushik Kasavajhala, Lauren Wickstrom, Kevin E Hauser, and Carlos Simmerling. ff14sb: improving the accuracy of protein side chain and backbone parameters from ff99sb. Journal of chemical theory and computation, 11(8):3696–3713, 2015

    2015

  116. [127]

    Development of force field parameters for the simulation of single-and double-stranded dna molecules and dna–protein complexes

    Maxwell R Tucker, Stefano Piana, Dazhi Tan, Michael V LeVine, and David E Shaw. Development of force field parameters for the simulation of single-and double-stranded dna molecules and dna–protein complexes. The Journal of Physical Chemistry B , 126(24):4442–4457, 2022

    2022

  117. [128]

    A general purpose model for the condensed phases of water: Tip4p/2005

    Jose LF Abascal and Carlos Vega. A general purpose model for the condensed phases of water: Tip4p/2005. The Journal of chemical physics , 123(23), 2005

    2005

  118. [129]

    All-atom empirical potential for molecular modeling and dynamics studies of proteins

    Alex D MacKerell Jr, Donald Bashford, MLDR Bellott, Roland Leslie Dunbrack Jr, Jeffrey D Evanseck, Martin J Field, Stefan Fischer, Jiali Gao, Houyang Guo, Sookhee Ha, et al. All-atom empirical potential for molecular modeling and dynamics studies of proteins. The journal of ph...

    1998

  119. [130]

    Slow dynamics in protein fluctuations revealed by time-structure based inde- pendent component analysis: the case of domain motions

    Yusuke Naritomi and Sotaro Fuchigami. Slow dynamics in protein fluctuations revealed by time-structure based inde- pendent component analysis: the case of domain motions. The Journal of chemical physics , 134(6), 2011

    2011

  120. [131]

    Kinetic distance and kinetic maps from molecular dynamics simulation

    Frank No´ e and Cecilia Clementi. Kinetic distance and kinetic maps from molecular dynamics simulation. Journal of chemical theory and computation , 11(10):5002–5011, 2015

    2015

  121. [132]

    Variational cross-validation of slow dynamical modes in molecular kinetics

    Robert T McGibbon and Vijay S Pande. Variational cross-validation of slow dynamical modes in molecular kinetics. The Journal of chemical physics , 142(12), 2015

    2015

  122. [133]

    Variational approach for learning markov processes from time series data.Journal of Nonlinear Science, 30(1):23–66, 2020

    Hao Wu and Frank No´ e. Variational approach for learning markov processes from time series data.Journal of Nonlinear Science, 30(1):23–66, 2020

    2020

  123. [134]

    Markov models of molecular kinetics: Generation and validation

    Jan-Hendrik Prinz, Hao Wu, Marco Sarich, Bettina Keller, Martin Senne, Martin Held, John D Chodera, Christof Sch¨ utte, and Frank No´ e. Markov models of molecular kinetics: Generation and validation. The Journal of chemical physics, 134(17), 2011

    2011

  124. [135]

    Identification of slow molecular order parameters for markov model construction

    Guillermo P´ erez-Hern´ andez, Fabian Paul, Toni Giorgino, Gianni De Fabritiis, and Frank No´ e. Identification of slow molecular order parameters for markov model construction. The Journal of chemical physics , 139(1), 2013

    2013

  125. [136]

    Animating rotation with quaternion curves

    Ken Shoemake. Animating rotation with quaternion curves. In Proceedings of the 12th annual conference on Computer graphics and interactive techniques , pages 245–254, 1985

    1985

  126. [137]

    Transition path theory

    Eric Vanden-Eijnden. Transition path theory. In Computer Simulations in Condensed Matter Systems: From Materials to Chemical Biology Volume 1 , pages 453–493. Springer, 2006

    2006

  127. [138]

    Constructing the equilibrium ensemble of folding pathways from short off-equilibrium simulations

    Frank No´ e, Christof Sch¨ utte, Eric Vanden-Eijnden, Lothar Reich, and Thomas R Weikl. Constructing the equilibrium ensemble of folding pathways from short off-equilibrium simulations. Proceedings of the National Academy of Sciences , 106(45):19011–19016, 2009

    2009

  128. [139]

    Estimating the sampling error: Distribution of transition matrices and functions of transition matrices for given trajectory data

    Philipp Metzner, Frank No´ e, and Christof Sch¨ utte. Estimating the sampling error: Distribution of transition matrices and functions of transition matrices for given trajectory data. Physical Review E—Statistical, Nonlinear, and Soft Matter Physics, 80(2):021106, 2009

    2009

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.