pith. sign in

arxiv: 2509.01038 · v2 · submitted 2025-09-01 · 🧬 q-bio.BM · cs.LG

Learning residue level protein dynamics with multiscale Gaussians

Mihir Bafna , Bowen Jing , Bonnie Berger This is my paper

Pith reviewed 2026-05-18 20:32 UTC · model grok-4.3

classification 🧬 q-bio.BM cs.LG
keywords protein dynamicsmultivariate Gaussiansresidue flexibilitycovariance matrixensemble generationstatic structureslightweight model
0
0 comments X

The pith

DynaProt predicts protein residue flexibility and dynamic couplings from static structures using multiscale Gaussians.

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

The paper introduces DynaProt as a lightweight framework to predict protein dynamics directly from static structures. It models dynamics as multivariate Gaussians at two scales: local per-residue anisotropy and pairwise couplings. This enables accurate prediction of flexibility measures like RMSF and reconstruction of covariance matrices for generating dynamic ensembles. The method uses far fewer parameters than prior approaches, making it scalable. Readers care because it offers a fast alternative to costly molecular dynamics simulations for understanding how proteins move and function.

Core claim

By framing protein dynamics through multivariate Gaussians, DynaProt estimates per-residue 3x3 covariance matrices for local flexibility and joint scalar covariances for pairwise dynamic coupling, allowing high-accuracy RMSF prediction and reasonable full covariance reconstruction from static structures alone.

What carries the argument

Multiscale Gaussians that separate marginal anisotropy for individual residues from scalar covariances encoding inter-residue dynamic couplings.

If this is right

  • High accuracy in residue-level flexibility prediction without running simulations.
  • Reasonable reconstruction of the full covariance matrix enables fast ensemble generation.
  • Uses orders of magnitude fewer parameters than previous methods.
  • Applicable directly to static structures for scalable analysis.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Could be combined with structure prediction models to generate dynamic ensembles for many proteins quickly.
  • May allow studying dynamics in contexts where simulations are infeasible due to size or time.
  • Potential to identify flexible regions important for function or binding more efficiently.

Load-bearing premise

That a Gaussian model trained on molecular dynamics data generalizes to predict biologically relevant dynamics for diverse new proteins.

What would settle it

Running molecular dynamics simulations on a test set of proteins and comparing the predicted RMSF values and reconstructed covariances to the simulation results.

Figures

Figures reproduced from arXiv: 2509.01038 by Bonnie Berger, Bowen Jing, Mihir Bafna.

Figure 1
Figure 1. Figure 1: Dynamics methods information content vs. efficiency. We introduce DynaProt, a lightweight, inter￾pretable, and expressive framework for predict￾ing protein dynamics through the lens of Gaussian distributions over structure (Section 2). Specifi￾cally, DynaProt predicts: (1) per-residue marginal anisotropy as 3 × 3 covariance matrices capturing local dynamics while encompassing RMSF, and (2) joint scalar N ×… view at source ↗
Figure 2
Figure 2. Figure 2: Overview of protein dynamics under Gaussian view. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: DynaProt architecture. assumption is motivated by the fact that the input structure usually corresponds to the experimentally determined (or AlphaFold-predicted) minimum energy conformation, and thus serves as a natural estimator of the ensemble mean. Consequently, the marginal prediction task reduces to learning the covariance matrices Σ (i) marginal alone. Marginal dynamics module. Recall that covariance… view at source ↗
Figure 4
Figure 4. Figure 4: DynaProt marginal Gaussian and residue coupling analysis. A. Renderings of pre￾dicted marginal Gaussians compared to ATLAS MD constructed Gaussians (mean symmetric KL divergence and RMWD are reported). B. Joint distribution (within 75th percentile) of DynaProt performance vs. (AFMD+T, NMA). C. Band-wise Pearson correlation between predicted and ground￾truth residue–residue coupling matrices as a function o… view at source ↗
Figure 5
Figure 5. Figure 5: DynaProt-M predicted residue Gaussians (ellipsoids) overlaid the apo form. 4.3.1 DynaProt zero-shot cryptic pocket discovery of Adenylosuccinate Synthetase Beyond accuracy, DynaProt-M’s marginals can also provide functional insight. Many proteins are considered to be undruggable as their apo form may not display a clear bind￾ing pocket. However, the druggable pocket may only become apparent after the drug … view at source ↗
Figure 6
Figure 6. Figure 6: Comparison of DynaProt generated ensemble vs. AFMD+T to ATLAS MD simulation (PDB 7qsu_A) overlaid on reference. RMSF Pearson correlation r and sample time reported. naProt consistently outperforms NMA across nearly all evaluations—except for transient contact prediction—particularly excelling in measures of local flexibility and pairwise distance preservation [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 8
Figure 8. Figure 8: DynaProt-M predicted RMSF correlations. Visualized test set examples of predicted RMSF per residue (derived from the predicted marginal Gaussians) compared to ground truth RMSF derived from MD trajectories. Pearson correlation coefficient (r) between predicted and ground truth RMSF is reported. 18 [PITH_FULL_IMAGE:figures/full_fig_p018_8.png] view at source ↗
read the original abstract

Many methods have been developed to predict static protein structures, however understanding the dynamics of protein structure is essential for elucidating biological function. While molecular dynamics (MD) simulations remain the in silico gold standard, its high computational cost limits scalability. We present DynaProt, a lightweight, SE(3)-invariant framework that predicts rich descriptors of protein dynamics directly from static structures. By casting the problem through the lens of multivariate Gaussians, DynaProt estimates dynamics at two complementary scales: (1) per-residue marginal anisotropy as $3 \times 3$ covariance matrices capturing local flexibility, and (2) joint scalar covariances encoding pairwise dynamic coupling across residues. From these dynamics outputs, DynaProt achieves high accuracy in predicting residue-level flexibility (RMSF) and, remarkably, enables reasonable reconstruction of the full covariance matrix for fast ensemble generation. Notably, it does so using orders of magnitude fewer parameters than prior methods. Our results highlight the potential of direct protein dynamics prediction as a scalable alternative to existing methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript introduces DynaProt, a lightweight SE(3)-invariant model that predicts protein dynamics directly from static structures by estimating per-residue 3×3 covariance matrices for local flexibility (anisotropy) and scalar pairwise covariances for inter-residue dynamic coupling. These outputs are assembled into a full covariance matrix to enable fast ensemble generation. The central claims are high accuracy on residue-level RMSF prediction and reasonable reconstruction of the full covariance, achieved with orders of magnitude fewer parameters than prior methods.

Significance. If the quantitative claims are substantiated, the work offers a scalable, low-parameter alternative to MD for generating protein ensembles and could accelerate dynamics-aware applications in structural biology. The multiscale Gaussian formulation and explicit SE(3) invariance are strengths that distinguish it from purely scalar flexibility predictors. The low parameter count is explicitly credited as a practical advantage.

major comments (3)
  1. [Abstract and §4] Abstract and §4 (Results): the claim of 'reasonable reconstruction of the full covariance matrix' for ensemble generation lacks reported quantitative metrics (Frobenius norm, eigenvalue spectrum match, or ensemble RMSD to held-out MD trajectories). Without these, it is impossible to evaluate whether the assembled 3N×3N matrix supports the headline claim beyond RMSF accuracy.
  2. [Methods (§3.2)] Methods (§3.2, covariance assembly): the per-residue 3×3 matrices and independently learned scalar pairwise terms are combined into a block matrix; no description is given of how positive-semidefiniteness is enforced (e.g., via Schur-complement constraints, projection, or regularization during training). This is load-bearing for the ensemble-generation claim and must be clarified with explicit checks or failure cases.
  3. [§4 and §5] §4 and §5: validation details (train/test splits, protein diversity, error bars, ablation of the scalar pairwise term) are not reported for the covariance reconstruction task. This weakens the generalization statement that the model captures 'biologically relevant dynamics across diverse proteins'.
minor comments (2)
  1. [Methods] Notation: the transition from per-residue 3×3 matrices to the full covariance should be given an explicit equation number for clarity.
  2. [Figures] Figure captions: add quantitative summary statistics (e.g., mean RMSF correlation) directly in the caption for quick reference.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for their thoughtful and constructive comments on our manuscript. We address each of the major comments point by point below. We have revised the manuscript to incorporate additional details and metrics as suggested.

read point-by-point responses

  1. Referee: [Abstract and §4] Abstract and §4 (Results): the claim of 'reasonable reconstruction of the full covariance matrix' for ensemble generation lacks reported quantitative metrics (Frobenius norm, eigenvalue spectrum match, or ensemble RMSD to held-out MD trajectories). Without these, it is impossible to evaluate whether the assembled 3N×3N matrix supports the headline claim beyond RMSF accuracy.

    Authors: We agree that quantitative metrics beyond RMSF would provide stronger support for the covariance reconstruction claim. In the revised version, we will report the Frobenius norm of the difference between predicted and MD-derived covariance matrices, compare eigenvalue spectra, and compute ensemble RMSDs for generated structures against held-out MD trajectories. These will be added to §4. revision: yes

  2. Referee: [Methods (§3.2)] Methods (§3.2, covariance assembly): the per-residue 3×3 matrices and independently learned scalar pairwise terms are combined into a block matrix; no description is given of how positive-semidefiniteness is enforced (e.g., via Schur-complement constraints, projection, or regularization during training). This is load-bearing for the ensemble-generation claim and must be clarified with explicit checks or failure cases.

    Authors: This is a valid point. The assembly process uses the fact that the per-residue 3x3 covariances are PSD by construction (as they are predicted as covariance matrices), and the scalar pairwise terms are incorporated in a way that maintains overall PSD through the multivariate Gaussian formulation and training regularization. However, we will expand §3.2 to explicitly describe the assembly procedure, how PSD is preserved, and include checks such as minimum eigenvalue distributions to confirm no violations occur in practice. revision: yes

  3. Referee: [§4 and §5] §4 and §5: validation details (train/test splits, protein diversity, error bars, ablation of the scalar pairwise term) are not reported for the covariance reconstruction task. This weakens the generalization statement that the model captures 'biologically relevant dynamics across diverse proteins'.

    Authors: We appreciate this feedback. While the main results in §4 focus on RMSF, the covariance task uses the same train/test splits and protein set as described in the methods. To address this, we will add in the revised §4 and §5: explicit mention of the splits for covariance evaluation, statistics on protein diversity (e.g., fold classes), error bars from cross-validation, and results from an ablation study where the scalar pairwise covariance term is removed to quantify its impact on full matrix reconstruction. revision: yes

Circularity Check

0 steps flagged

No significant circularity: predictions learned from external MD data

full rationale

The paper frames DynaProt as a supervised learning model trained on molecular dynamics trajectories to output per-residue 3x3 covariance matrices and scalar inter-residue covariances from static structures. These outputs are not defined in terms of each other or the target RMSF values inside the same equations; the full covariance reconstruction is described as a post-hoc assembly enabled by the learned descriptors rather than a quantity forced by construction or by fitting a subset and renaming it. No self-citation load-bearing steps, uniqueness theorems imported from the authors' prior work, or ansatzes smuggled via citation appear in the derivation. The model is presented as generalizing from external data, so the claimed predictions retain independent content and the chain is self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on the domain assumption that Gaussian distributions are sufficient to represent essential protein dynamics at the residue scale; no free parameters or invented entities are explicitly introduced in the abstract.

axioms (1)
  • domain assumption Protein dynamics can be adequately represented by multivariate Gaussian distributions at per-residue and pairwise scales.
    The problem is cast through the lens of multivariate Gaussians to estimate marginal anisotropy and joint covariances.

pith-pipeline@v0.9.0 · 5707 in / 1263 out tokens · 37035 ms · 2026-05-18T20:32:44.699209+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Motion-Enabled Tomography via Gaussian Mixture Models

    math.NA 2026-05 unverdicted novelty 7.0

    A parametric GMM model for motion-enabled tomography that decouples reconstruction into sub-problems and tests on 2D simulations of intersecting trajectories.

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages · cited by 1 Pith paper · 1 internal anchor

  1. [1]

    Accurate structure prediction of biomolecular interactions with alphafold 3

    Josh Abramson, Jonas Adler, Jack Dunger, Richard Evans, Tim Green, Alexander Pritzel, Olaf Ronneberger, Lindsay Willmore, Andrew J Ballard, Joshua Bambrick, et al. Accurate structure prediction of biomolecular interactions with alphafold 3. Nature, 630 0 (8016): 0 493--500, 2024

    work page 2024

  2. [2]

    On the Bures--Wasserstein distance between positive definite matrices

    Rajendra Bhatia, Tanvi Jain, and Yongdo Lim. On the Bures--Wasserstein distance between positive definite matrices . Expositiones Mathematicae , 37 0 (2): 0 165--191, 2019

    work page 2019

  3. [3]

    Protein data bank (pdb): the single global macromolecular structure archive

    Stephen K Burley, Helen M Berman, Gerard J Kleywegt, John L Markley, Haruki Nakamura, and Sameer Velankar. Protein data bank (pdb): the single global macromolecular structure archive. Protein crystallography: methods and protocols, pages 627--641, 2017

    work page 2017

  4. [4]

    Insights from molecular dynamics simulations for computational protein design

    Matthew Carter Childers and Valerie Daggett. Insights from molecular dynamics simulations for computational protein design. Molecular systems design & engineering, 2 0 (1): 0 9--33, 2017

    work page 2017

  5. [5]

    Exploring cryptic pockets formation in targets of pharmaceutical interest with swish

    Federico Comitani and Francesco Luigi Gervasio. Exploring cryptic pockets formation in targets of pharmaceutical interest with swish. Journal of chemical theory and computation, 14 0 (6): 0 3321--3331, 2018

    work page 2018

  6. [6]

    Normal mode analysis: theory and applications to biological and chemical systems

    Qiang Cui and Ivet Bahar. Normal mode analysis: theory and applications to biological and chemical systems. CRC press, 2005

    work page 2005

  7. [7]

    Sampling alternative conformational states of transporters and receptors with alphafold2

    Diego Del Alamo, Davide Sala, Hassane S Mchaourab, and Jens Meiler. Sampling alternative conformational states of transporters and receptors with alphafold2. Elife, 11: 0 e75751, 2022

    work page 2022

  8. [8]

    Deep sparse rectifier neural networks

    Xavier Glorot, Antoine Bordes, and Yoshua Bengio. Deep sparse rectifier neural networks . In Proceedings of the fourteenth international conference on artificial intelligence and statistics , pages 315--323. JMLR Workshop and Conference Proceedings, 2011

    work page 2011

  9. [9]

    Deep learning--guided design of dynamic proteins

    Amy B Guo, Deniz Akpinaroglu, Christina A Stephens, Michael Grabe, Colin A Smith, Mark JS Kelly, and Tanja Kortemme. Deep learning--guided design of dynamic proteins. Science, 388 0 (6749): 0 eadr7094, 2025

    work page 2025

  10. [10]

    Molecular dynamics simulation for all

    Scott A Hollingsworth and Ron O Dror. Molecular dynamics simulation for all. Neuron, 99 0 (6): 0 1129--1143, 2018

    work page 2018

  11. [11]

    Cryptic pocket formation underlies allosteric modulator selectivity at muscarinic gpcrs

    Scott A Hollingsworth, Brendan Kelly, Celine Valant, Jordan Arthur Michaelis, Olivia Mastromihalis, Geoff Thompson, AJ Venkatakrishnan, Samuel Hertig, Peter J Scammells, Patrick M Sexton, et al. Cryptic pocket formation underlies allosteric modulator selectivity at muscarinic gpcrs. Nature Communications, 10 0 (1): 0 3289, 2019

    work page 2019

  12. [12]

    Seqdance: A protein language model for representing protein dynamic properties

    Chao Hou and Yufeng Shen. Seqdance: A protein language model for representing protein dynamic properties. bioRxiv, 2024

    work page 2024

  13. [13]

    Log-euclidean metric learning on symmetric positive definite manifold with application to image set classification

    Zhiwu Huang, Ruiping Wang, Shiguang Shan, Xianqiu Li, and Xilin Chen. Log-euclidean metric learning on symmetric positive definite manifold with application to image set classification . In International conference on machine learning , pages 720--729. PMLR, 2015

    work page 2015

  14. [14]

    The theory of probability

    Harold Jeffreys. The theory of probability. OuP Oxford, 1998

    work page 1998

  15. [15]

    Alphafold meets flow matching for generating protein ensembles, 2024

    Bowen Jing, Bonnie Berger, and Tommi Jaakkola. AlphaFold meets flow matching for generating protein ensembles . arXiv preprint arXiv:2402.04845, 2024

    work page arXiv 2024

  16. [16]

    Highly accurate protein structure prediction with AlphaFold

    John Jumper, Richard Evans, Alexander Pritzel, Tim Green, Michael Figurnov, Olaf Ronneberger, Kathryn Tunyasuvunakool, Russ Bates, Augustin Z \' dek, Anna Potapenko, et al. Highly accurate protein structure prediction with AlphaFold . Nature , 596 0 (7873): 0 583--589, 2021

    work page 2021

  17. [17]

    Learning to engineer protein flexibility

    Petr Kouba, Joan Planas-Iglesias, Jiri Damborsky, Jiri Sedlar, Stanislav Mazurenko, and Josef Sivic. Learning to engineer protein flexibility. arXiv preprint arXiv:2412.18275, 2024

    work page arXiv 2024

  18. [18]

    On information and sufficiency

    Solomon Kullback and Richard A Leibler. On information and sufficiency. The annals of mathematical statistics, 22 0 (1): 0 79--86, 1951

    work page 1951

  19. [19]

    Scalable emulation of protein equilibrium ensembles with generative deep learning

    Sarah Lewis, Tim Hempel, Jos \'e Jim \'e nez-Luna, Michael Gastegger, Yu Xie, Andrew YK Foong, Victor Garc \' a Satorras, Osama Abdin, Bastiaan S Veeling, Iryna Zaporozhets, et al. Scalable emulation of protein equilibrium ensembles with generative deep learning. bioRxiv, pages 2024--12, 2024

    work page 2024

  20. [20]

    Seamoon: Prediction of molecular motions based on language models

    Valentin Lombard, Dan Timsit, Sergei Grudinin, and Elodie Laine. Seamoon: Prediction of molecular motions based on language models. bioRxiv, pages 2024--09, 2024

    work page 2024

  21. [21]

    Structure language models for protein conformation generation

    Jiarui Lu, Xiaoyin Chen, Stephen Zhewen Lu, Chence Shi, Hongyu Guo, Yoshua Bengio, and Jian Tang. Structure language models for protein conformation generation. arXiv preprint arXiv:2410.18403, 2024

    work page arXiv 2024

  22. [22]

    Usefulness and limitations of normal mode analysis in modeling dynamics of biomolecular complexes

    Jianpeng Ma. Usefulness and limitations of normal mode analysis in modeling dynamics of biomolecular complexes. Structure, 13 0 (3): 0 373--380, 2005

    work page 2005

  23. [23]

    Protein dynamics

    JA McCammon. Protein dynamics. Reports on Progress in Physics, 47 0 (1): 0 1, 1984

    work page 1984

  24. [24]

    Lotthammer, Felipe Oviedo, Juan Lavista Ferres, and Gregory R

    Artur Meller, Michael Ward, Jonathan Borowsky, Meghana Kshirsagar, Jeffrey M. Lotthammer, Felipe Oviedo, Juan Lavista Ferres, and Gregory R. Bowman. Predicting locations of cryptic pockets from single protein structures using the PocketMiner graph neural network. 14 0 (1): 0 1177. ISSN 2041-1723. doi:10.1038/s41467-023-36699-3. URL https://doi.org/10.1038...

    work page doi:10.1038/s41467-023-36699-3 2041

  25. [25]

    undruggable

    Minyue Mou, Weicheng Yang, Guangyi Huang, Xiaoyan Yang, Xiao Zhang, Wasala Mudiyanselage Wishwajith Wickramabahu Kandegama, Charles R Ashby Jr, Gefei Hao, and Yangyang Gao. The discovery of cryptic pockets increases the druggability of “undruggable” proteins. Medicinal Research Reviews, 2025

    work page 2025

  26. [26]

    Fast procedure for reconstruction of full-atom protein models from reduced representations

    Piotr Rotkiewicz and Jeffrey Skolnick. Fast procedure for reconstruction of full-atom protein models from reduced representations. Journal of computational chemistry, 29 0 (9): 0 1460--1465, 2008

    work page 2008

  27. [27]

    Atomic-level characterization of the structural dynamics of proteins

    David E Shaw, Paul Maragakis, Kresten Lindorff-Larsen, Stefano Piana, Ron O Dror, Michael P Eastwood, Joseph A Bank, John M Jumper, John K Salmon, Yibing Shan, et al. Atomic-level characterization of the structural dynamics of proteins. Science, 330 0 (6002): 0 341--346, 2010

    work page 2010

  28. [28]

    Normal mode analysis for proteins

    Lars Skjaerven, Siv M Hollup, and Nathalie Reuter. Normal mode analysis for proteins. Journal of Molecular Structure: THEOCHEM, 898 0 (1-3): 0 42--48, 2009

    work page 2009

  29. [29]

    Speach\_af: Sampling protein ensembles and conformational heterogeneity with alphafold2

    Richard A Stein and Hassane S Mchaourab. Speach\_af: Sampling protein ensembles and conformational heterogeneity with alphafold2. PLOS Computational Biology, 18 0 (8): 0 e1010483, 2022

    work page 2022

  30. [30]

    ATLAS: protein flexibility description from atomistic molecular dynamics simulations

    Yann Vander Meersche, Gabriel Cretin, Aria Gheeraert, Jean-Christophe Gelly, and Tatiana Galochkina. ATLAS: protein flexibility description from atomistic molecular dynamics simulations . Nucleic Acids Research, 52 0 (D1): 0 D384--D392, 2024

    work page 2024

  31. [31]

    Riemannian Metric Learning for Symmetric Positive Definite Matrices

    Raviteja Vemulapalli and David W Jacobs. Riemannian metric learning for symmetric positive definite matrices . arXiv preprint arXiv:1501.02393, 2015

    work page internal anchor Pith review Pith/arXiv arXiv 2015

  32. [32]

    Protein conformation generation via force-guided se (3) diffusion models

    Yan Wang, Lihao Wang, Yuning Shen, Yiqun Wang, Huizhuo Yuan, Yue Wu, and Quanquan Gu. Protein conformation generation via force-guided se (3) diffusion models. arXiv preprint arXiv:2403.14088, 2024

    work page arXiv 2024

  33. [33]

    Predicting multiple conformations via sequence clustering and alphafold2

    Hannah K Wayment-Steele, Adedolapo Ojoawo, Renee Otten, Julia M Apitz, Warintra Pitsawong, Marc H \"o mberger, Sergey Ovchinnikov, Lucy Colwell, and Dorothee Kern. Predicting multiple conformations via sequence clustering and alphafold2. Nature, 625 0 (7996): 0 832--839, 2024

    work page 2024

  34. [34]

    Learning millisecond protein dynamics from what is missing in nmr spectra

    Hannah K Wayment-Steele, Gina El Nesr, Ramith Hettiarachchi, Hasindu Kariyawasam, Sergey Ovchinnikov, and Dorothee Kern. Learning millisecond protein dynamics from what is missing in nmr spectra. bioRxiv, pages 2025--03, 2025

    work page 2025

  35. [35]

    Propagating conformational changes over long (and short) distances in proteins

    Edward W Yu and Daniel E Koshland Jr. Propagating conformational changes over long (and short) distances in proteins. Proceedings of the National Academy of Sciences, 98 0 (17): 0 9517--9520, 2001

    work page 2001

  36. [36]

    G protein-coupled receptors (gpcrs): advances in structures, mechanisms and drug discovery

    Mingyang Zhang, Ting Chen, Xun Lu, Xiaobing Lan, Ziqiang Chen, and Shaoyong Lu. G protein-coupled receptors (gpcrs): advances in structures, mechanisms and drug discovery. Signal Transduction and Targeted Therapy, 9 0 (1): 0 88, 2024

    work page 2024

  37. [37]

    Prody 2.0: increased scale and scope after 10 years of protein dynamics modelling with python

    She Zhang, James M Krieger, Yan Zhang, Cihan Kaya, Burak Kaynak, Karolina Mikulska-Ruminska, Pemra Doruker, Hongchun Li, and Ivet Bahar. Prody 2.0: increased scale and scope after 10 years of protein dynamics modelling with python. Bioinformatics, 37 0 (20): 0 3657--3659, 2021

    work page 2021

  38. [38]

    Activation and friction in enzymatic loop opening and closing dynamics

    Kirill Zinovjev, Paul Gu \'e non, Carlos A Ramos-Guzm \'a n, J Javier Ruiz-Pern \' a, Damien Laage, and I \ n aki Tu \ n \'o n. Activation and friction in enzymatic loop opening and closing dynamics. Nature Communications, 15 0 (1): 0 2490, 2024

    work page 2024