Pith. sign in

REVIEW 1 major objections 5 minor 73 references

Active Gaussian Network Model: a non-equilibrium description of protein fluctuations and allosteric behavior

T0 review · 1 major / 5 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read The paper claims that adding a simple colored-noise, non-equilibrium bath to the Gaussian Network Model preserves the model's solvability while changing the timescales and memory of protein fluctuations, and that causal indicators on this…

desk verdict The aGNM model and transfer entropy are worth keeping; the response function that drives the central causal claims is computed with a formula that fails when active variables are hidden. read the letter →

arxiv 2505.24855 v1 pith:4LRLWWFL submitted 2025-05-30 cond-mat.stat-mech cond-mat.soft

Giulio Costantini , Lorenzo Caprini , Umberto Marini Bettolo Marconi , Fabio Cecconi This is my paper · ORCID
classification cond-mat.stat-mechcond-mat.soft
keywords activeGaussiannetworkmodelallosteryPDZ2domainOrnstein-Uhlenbeckprocessestransferentropyresponsefunctionsnon-equilibriumfluctuations
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

This paper introduces the active Gaussian Network Model (aGNM), a minimal extension of the standard Gaussian Network Model in which each residue is stirred by an Ornstein-Uhlenbeck colored noise instead of a thermal bath alone. The model remains analytically solvable through mode diagonalization and reproduces experimental B-factors of the PDZ2 domain while amplifying flexibility in flexible regions. Using response functions and transfer entropy as causal indicators, the authors argue that in PDZ2 the beta2 strand acts as a driver of the allosteric signal, the beta2-H1-beta3 loop is the central hub, and helix H3 plays only a marginal role. They conclude that deviations from purely thermal fluctuations can significantly influence allosteric communication by introducing distinct timescales and memory effects, which matters when allosteric responses unfold on timescales that do not allow relaxation to equilibrium.

What carries the argument

The central object is the active Gaussian Network Model, a multivariate Ornstein-Uhlenbeck process coupling residue displacements $x_i$ to active velocities $w_i$ that relax on a memory timescale $\tau_a$; writing the dynamics in the eigenbasis of the Kirchhoff matrix $K$ reduces the problem to independent modes with effective relaxation rates $\mu_k$. From that solution the paper derives explicit formulas for the correlation matrix, the entropy production rate, the response function $R(t) = C(t)C^+(0)$, and the Gaussian transfer entropy. The response formula and the net entropy transfer are the two causal indicators that carry the allosteric conclusions, and their explicit modal expressions are what keep the model analytically tractable.

What would settle it

Simulate the aGNM equations with an explicit instantaneous displacement of residue 21 and measure the average response of each residue from the simulated ensemble, then compare with Eq. (19); if the simulated responses deviate systematically from the formula as $\tau_a$ increases, the response-based allosteric network is an artifact of the x-only formula.

Watch

Extended reading notes

Core claim

The central claim is that allostery can be read as a causal process in a non-equilibrium harmonic network: replacing the thermal bath of the GNM with active Ornstein-Uhlenbeck noise leaves the native-fold topology as the determinant of fluctuation patterns, but alters the statistical weight of the normal modes, slows relaxation, and adds memory. On the PDZ2 domain, the resulting response and transfer-entropy profiles indicate that a perturbation at the binding groove propagates mainly through the beta2 strand, that the beta2-H1-beta3 loop is the most sensitive relay, and that helix H3 is a comparatively marginal sender. The paper states this as the finding that deviations from purely thermal fluctuations can significantly influence allosteric communication by introducing distinct timescales and memory effects, particularly when the allosteric response occurs on timescales incompatible with equilibrium relaxation. If correct, the paper establishes that a simple, analytically solvable non-equilibrium elastic network can reproduce and explain known allosteric communication patterns in a protein.

Load-bearing premise

The causal-response analysis assumes that the response of residue displacements to an initial perturbation is completely determined by the displacement-only correlation matrix $C(t) C^+(0)$, ignoring the hidden active-noise variables that the model itself introduces.

Editorial extensions

If this is right

  • Non-equilibrium fluctuations can change the ranking of which structural elements mediate allostery, so equilibrium elastic-network analyses may systematically undervalue roles like the beta2 strand.
  • The aGNM offers a low-cost, analytically solvable route to identify candidate allosteric pathways in any protein with a known contact map, without molecular-dynamics sampling.
  • Memory effects make self-response decay non-exponential, implying that allosteric propagation should be characterized by timescales rather than static correlation patterns alone.
  • The causal indicators select specific secondary-structure elements, generating targeted predictions about which mutations should perturb PDZ2 allostery.
  • Differences between active and thermal correlation patterns localize near native contacts and the binding groove, suggesting that activity amplifies already-coordinated motions rather than creating new ones.

Reading between the lines

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

  • A natural next test is to apply the aGNM to other allosteric proteins and compare each predicted driver region against experimentally mapped allosteric sites; the model would predict that the dominant sender sits at the edge of the ligand-binding site rather than at the active site.
  • Because the x-only response formula is used, a direct check on simulated aGNM trajectories with explicit perturbations would clarify whether the causal network survives when hidden active variables are accounted for.
  • If the causal network holds, the analytic formulas make the aGNM a fast screening tool for ranking allosteric mutations in any protein with a known native contact map.
  • The paper's reliance on native topology alone implicitly suggests that fold geometry, not sequence, sets the backbone of allosteric communication; testing this by randomizing sequences on the same fold would separate the two.
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

1 major / 5 minor

Summary. The manuscript introduces the active Gaussian Network Model (aGNM), in which the standard GNM harmonic network for Cα fluctuations is supplemented by an Ornstein-Uhlenbeck 'active velocity' with persistence time τ_a and strength T_a. The authors derive exact stationary correlation functions, B-factors, and a positive entropy production rate, and they compare the predicted B-factors with experimental data for the PDZ2 domain (PDB 3LNX). They then use two causal indicators, response functions and transfer entropy, to infer allosteric communication pathways in PDZ2, concluding that the β2 strand acts as a driver, the β2-H1-β3 loop is a central hub, and helix H3 plays a marginal role. The central claim is that non-thermal, memory-carrying fluctuations can substantially modify allosteric communication by introducing distinct timescales and memory effects.

Significance. If the analysis were correct, the aGNM would be a valuable analytically solvable platform for studying nonequilibrium effects in protein allostery. The model is simple, and the correlation and entropy-production derivations in Appendices A and B are transparent and self-consistent; the transfer-entropy formulas are explicit, and the allosteric conclusions are not fitted to the experimental allosteric data, so the TE-based part is not circular. However, the response-function analysis, which is one of the two causal pillars and the source of the block-response network in Fig. 8(a), is invalid for the stated perturbation protocol. The allosteric conclusions are therefore only partially supported, and the significance of the paper is conditional on the response section being corrected or removed.

major comments (1)
  1. [Sec. V B, Eq. (19)] The quantity computed in Eq. (19) is not the response defined in this section. The definition R_ij(t)=δ⟨x_j(t)⟩/δx_i(0) with the perturbed average taken over the unperturbed stationary distribution, and with the active variables w explicitly excluded from the response, is the physical response to an initial shift of x with w drawn from its stationary mean ⟨w⟩=0. For the linear mean dynamics γ d⟨x⟩/dt = -gK⟨x⟩ + γ⟨w⟩ and d⟨w⟩/dt = -⟨w⟩/τ_a, this gives ⟨w(t)⟩=0 and hence R_ij(t)=[exp(-(g/γ)K t)]_{ij} = Σ_k u_i(k) e^{-μ_k t} u_j(k), independent of T_a and τ_a. The formula R(t)=C(t)C^+(0) is the optimal linear predictor (regression) of x(t) from x(0), not the response under this protocol; it would become the response only if w(0) were prepared with its conditional distribution given the shifted x(0). Therefore the non-exponential 'memory' decay shown in the inset of Fig. 5, the response profiles of Fig. 6, and the block-response network of Fig. 8(a) are artifacts of applying the regression formula to a system with hidden active degrees of freedom. The correlation and transfer-entropy analyses are unaffected, but the response-based conclusion that β2 is a driver and H3 is marginal in allosteric communication is unsupported. The authors should either recompute the response under the stated protocol (which will remove the τ_a and T_a dependence) or clearly re-label the quantity and revise the abstract and conclusions accordingly.
minor comments (5)
  1. [Fig. 5 and Sec. V B] The pair 21–38 is described both as a direct native contact ('in direct harmonic links') and as a non-contact pair connected by a short path; the caption and the surrounding text should be made consistent.
  2. [Throughout] The PDB identifier appears as both '3LNX' (text) and '3NLX' (Fig. 1), and the protein name as both 'hPT1E' (abstract) and 'hPTP1E' (Section II); these should be standardized.
  3. [Eq. (18)] The small-τ_a expansion appears to set the harmonic scale g to unity, since the coefficients of the δ_ij and K^m terms differ from a direct expansion of Eq. (16) by powers of g; please state the value of g or the reduced units used in the numerics.
  4. [Sec. V C, Eq. (21)] The quantity in Eq. (21) conditions only on the single past values x_i(0) and x_j(0). Since the x-process of the aGNM is non-Markovian due to the hidden active variables, this is a lag-specific conditional mutual information rather than the full Schreiber transfer entropy with histories; the text should qualify the causal interpretation accordingly.
  5. [Sec. V C, Eq. (25)] The sentence following Eq. (25) contains a sign/logic typo: if δTE_i,j>0 and residue i gives entropy, then i is the driver, not j; please correct.

Circularity Check

1 steps flagged · score 4.0 of 10

The aGNM correlation and entropy-production derivations are self-contained and the allosteric conclusions are not fitted to target data, but the load-bearing response analysis (Eq. 19) is imported from self-authored references and reduces the claimed causal response to a correlation regression, so the response-based portion of the central claim is not independent.

  1. self citation load bearing [Section V B, Eq. (19) and the paragraph introducing R(t)=C(t)C+(0)]
    "For linear stochastic systems like the aGNM, the response functions between residues can be obtained from the knowledge of correlations via the simplified formula [52, 64] R(t) = C(t)C +(0) with C(t) being the time-dependent covariance matrix restricted to residues only, C(t) = ⟨x(t)x⊤(0)⟩, and C +(0) is its pseudo-inverse at time, t= 0. It is important to note that this response is restricted to the residue displacements x(t), excluding the active-noise variables w(t) introduced in the aGNM dynamics (5)."

    Response is defined as δ⟨x_j(t)⟩/δx_i(0) with unperturbed stationary initial conditions, but Eq. (19) is evaluated as C(t)C^+(0). For the full aGNM state, d⟨w⟩/dt = −⟨w⟩/τ_a, so with w unperturbed the exact response is exp(−(g/γ)K t), independent of T_a and τ_a. Eq. (19) is instead the optimal linear regression of x(t) on x(0). The non-exponential decay (Fig. 5 inset), the 'marginal H3' profile, and the Fig. 8(a) network are thus generated by a self-cited correlation formula, not by the defined perturbation protocol. This is load-bearing for the claim that β2 drives allostery while H3 is marginal.

full rationale

The mathematical core of the aGNM is not circular: Eqs. (4)-(5) define a solvable OU process, Appendix B derives the correlation functions from those dynamics, and the entropy production in Appendix A is a path-integral calculation. The B-factor comparison uses external PDB data and only a mean rescaling, with τ_a and T_a set by hand rather than optimized to the allosteric target, so no fitted parameter is being renamed as a prediction. The equal-time correlation and transfer-entropy analyses are also internally derived from the model's Gaussian statistics. The one load-bearing self-citation is the response formula R(t)=C(t)C^+(0), imported from Refs. [52,64] (with overlapping authorship). Under the response protocol stated in the paper, that formula does not give the physical response when the active variables w are hidden and unperturbed; it gives the correlation-based regression, so the non-equilibrium 'memory' response and the response-based causal network reduce to a re-expression of correlations. Because the transfer-entropy and correlation pillars remain independent, the circularity is partial rather than total.

Assumptions & free parameters 6 free parameters · 6 assumptions · 1 invented entities

The model introduces several hand-chosen parameters (T_a, tau_a, A0, cutoff) that shape the results, though they are not fitted to the allosteric output. The most consequential assumption is the response formula for hidden active variables, which is load-bearing for the causal interpretation. The active velocity is a model device rather than an observed entity.

free parameters (6)
  • T_a (active temperature strength) = T_a = 20 T_0, with T_0 = 1
    Set by hand to explore the non-equilibrium regime; controls the amplitude of active fluctuations; no systematic scan or data fit.
  • tau_a (active persistence time) = tau_a = 0.1 for main results; varied 0.01 to 1.0
    Set by hand; controls the memory of the active noise; main results use 0.1 in dimensionless units.
  • A0 (backbone spring strength) = A0 = 10 A_ij, with A_ij = 1
    Chosen to distinguish backbone links from other contacts; a standard convention in GNM, but still an arbitrary choice.
  • r_c (contact cutoff) = r_c = 5 Å
    Used to define heavy-atom contacts from the PDB structure; standard but arbitrary and affects the Kirchhoff matrix.
  • g (harmonic potential scale) = unspecified; cancels in normalized B-factors
    Sets the energy scale; not fitted because the B-factors are rescaled by their mean, but it affects dynamical timescales through mu_k = g lambda_k / gamma.
  • gamma (friction coefficient) = unspecified; sets the time unit
    An overall inverse time scale; figures plot gamma t, so only relative values matter.
assumptions (6)
  • domain assumption A protein's equilibrium fluctuations can be represented by a scalar Gaussian network of C-alpha beads connected by harmonic springs.
    Section III builds the aGNM on the standard GNM assumption that native topology and harmonic interactions capture protein flexibility.
  • domain assumption The heavy-atom contact map with cutoff 5 Å from PDB 3LNX captures the relevant allosteric connectivity.
    Section III and Figure 1(c); the Kirchhoff matrix is built from this map, so all results inherit this structural assumption.
  • ad hoc to paper Non-equilibrium environmental fluctuations can be modeled by an Ornstein-Uhlenbeck active velocity with a single persistence time tau_a, independent of thermal noise.
    Equations (4) and (5); the paper justifies this by analogy to active matter, but it is a modeling choice not derived from protein physics.
  • ad hoc to paper For a linear non-equilibrium system, the response of residue displacements to an initial perturbation is given by R(t)=C(t)C^+(0) using only the displacement correlation matrix.
    Section V B, after Eq. (19). This is a major assumption when active variables are hidden; the paper cites refs. 52 and 64 but does not prove validity for non-Markovian hidden-variable dynamics.
  • standard math The Gaussian transfer entropy formula (Eq. 22) is valid for the stationary aGNM process.
    The aGNM is a multivariate Gaussian process, so the Gaussian TE expression from refs. 52, 67, 69 applies.
  • domain assumption The active velocity w_i is odd under time reversal for the entropy production calculation.
    Appendix A, Eq. (A4) and surrounding text; the paper notes this parity choice is debated in the active matter literature.
invented entities (1)
  • Active velocity w_i
    purpose: A hidden memory variable attached to each residue that produces non-equilibrium, colored-noise fluctuations.
    Mathematical construct inspired by active matter; it is not claimed to be a physical particle or force, and no independent experimental handle is offered. It drives the entropy production and memory effects in the model.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Active Gaussian Network Model: a non-equilibrium description of protein fluctuations and allosteric behavior." pith.science (2026). https://pith.science/paper/4LRLWWFL

@misc{pith2026250524855,
  author       = {Pith},
  title        = {Pith review of: Active Gaussian Network Model: a non-equilibrium description of protein fluctuations and allosteric behavior},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/4LRLWWFL}},
  note         = {Machine review of arXiv:2505.24855}
}
read the original abstract

Understanding the link between structure and function in proteins is fundamental in molecular biology and proteomics. A central question in this context is whether allostery - where the binding of a molecule at one site affects the activity of a distant site - emerges as a further manifestation of the intricate interplay between structure, function, and intrinsic dynamics. This study explores how allosteric regulation is modified when intrinsic protein dynamics operate under out-of-equilibrium conditions. To this purpose, we introduce a simple nonequilibrium model of protein dynamics, inspired by active matter systems, by generalizing the widely employed Gaussian Network Model (GNM) to incorporate non-thermal effects. Our approach underscores the advantage of framing allostery as a causal process by using, as a benchmark system, the second PDZ domain of the human phosphatase hPT1E that mediates protein-protein interactions. We employ causal indicators, such as response functions and transfer entropy, to identify the network of PDZ2 residues through which the allosteric signal propagates across the protein structure. These indicators reveal specific regions that align well with experimental observations. Furthermore, our results suggest that deviations from purely thermal fluctuations can significantly influence allosteric communication by introducing distinct timescales and memory effects. This influence is particularly relevant when the allosteric response unfolds on timescales incompatible with relaxation to equilibrium. Accordingly, non-thermal fluctuations may become essential for accurately describing protein responses to ligand binding and developing a comprehensive understanding of allosteric regulation.

Figures

Figures reproduced from arXiv: 2505.24855 by the authors.

Figure 1
Figure 1. FIG. 1. Structure and topology of the PDZ2 analyzed in this work (PDBid: 3NLX). Panel (a): [PITH_FULL_IMAGE:figures/full_fig_p006_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. Performance check of aGNM. Panel (a) comparison of: i) thermal B-factors, obtained at [PITH_FULL_IMAGE:figures/full_fig_p014_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. Color maps of the difference, [PITH_FULL_IMAGE:figures/full_fig_p016_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: FIG. 4. The correlation profiles, [PITH_FULL_IMAGE:figures/full_fig_p018_4.png]
Figure 5
Figure 5. Figure 5: FIG. 5. Time courses of responses [PITH_FULL_IMAGE:figures/full_fig_p020_5.png]
Figure 6
Figure 6. Figure 6: FIG. 6. Response profiles [PITH_FULL_IMAGE:figures/full_fig_p021_6.png]
Figure 7
Figure 7. Figure 7: FIG. 7. Net entropy transfer (NET) profiles referring to residue 21, panel (a) and residue 76 panel [PITH_FULL_IMAGE:figures/full_fig_p025_7.png]
Figure 8
Figure 8. Figure 8: FIG. 8. Causal networks emerging from the analysis of causal indicators (a) block response (b) [PITH_FULL_IMAGE:figures/full_fig_p026_8.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

73 extracted references · 72 canonical work pages

  1. [1]

    Lodish, A

    H. Lodish, A. Berk, S. Zipursky, P. Matsudaira, B. D. and J. Darnell,Molecular Cell Biology, 4th Edition, W.H.Freeman & Co Ltd, USA, 2002

    work page 2002

  2. [2]

    Smock and L

    R. Smock and L. Gierasch,Science, 2009,324, 198–203. 32

    work page 2009

  3. [3]

    Kubitzki, B

    M. Kubitzki, B. de Groot and D. Seeliger, inProtein Dynamics: From Structure to Function, Springer Netherlands, Dordrecht, 2017, pp. 393–425

    work page 2017

  4. [4]

    Lesk,Introduction to protein science: architecture, function, and genomics, 3rd edition, Oxford University Press, USA, 2016

    A. Lesk,Introduction to protein science: architecture, function, and genomics, 3rd edition, Oxford University Press, USA, 2016

    work page 2016

  5. [5]

    McLaughlin Jr, F

    R. McLaughlin Jr, F. Poelwijk, A. Raman, W. Gosal and R. Ranganathan,Nature, 2012,491, 138–142

    work page 2012

  6. [6]

    Kessel and N

    A. Kessel and N. Ben-Tal,Introduction to proteins: structure, function, and motion, Chapman and Hall/CRC, New York USA, 2018

    work page 2018

  7. [7]

    Bahar, T

    I. Bahar, T. Lezon, L.-W. Yang and E. Eyal,Annu. Rev. Biophys., 2010,39, 23–42

    work page 2010

  8. [8]

    Erman,Phys

    B. Erman,Phys. Bio., 2025,22, 026003

    work page 2025

Show all 73 references
  1. [9]

    Hub and B

    J. Hub and B. de Groot,PLoS Comput. Biol., 2009,5, e1000480

    2009

  2. [10]

    E. C. Dykeman and O. F. Sankey,J. Phys. Condens. Matter, 2010,22, 423202

    2010

  3. [11]

    Frauenfelder, B

    H. Frauenfelder, B. H. McMahon, R. H. Austin, K. Chu and J. T. Groves,Proc. Natl. Acad. Sci., 2001,98, 2370–2374

    2001

  4. [12]

    Liu and R

    J. Liu and R. Nussinov,PLoS Comput. Biol., 2016,12, e1004966

    2016

  5. [13]

    Thirumalai, C

    D. Thirumalai, C. Hyeon, P. Zhuravlev and G. Lorimer,Chem. Rev., 2019,119, 6788–6821

    2019

  6. [14]

    Ribeiro and V

    A.-A. Ribeiro and V. Ortiz,Chem. Rev., 2016,116, 6488–6502

    2016

  7. [15]

    Monod and F

    J. Monod and F. Jacob, Cold Spring Harbor symposia on quantitative biology, 1961, pp. 389–401

    1961

  8. [16]

    Guarnera and I

    E. Guarnera and I. N. Berezovsky,Curr. Opin. Struct. Biol., 2016,37, 1–8

    2016

  9. [17]

    Cooper and D

    A. Cooper and D. Dryden,Eur. Biophys. J., 1984,11, 103–109. [18]Allostery: Methods and Protocols, ed. A. Fenton, Humana Totowa, NJ, New York, NY, 2012, vol. 796

    1984

  10. [19]

    Guarnera and I

    E. Guarnera and I. N. Berezovsky,PLoS Comput. Biol., 2016,12, e1004678

    2016

  11. [20]

    Tirion,Phys

    M. Tirion,Phys. Rev. Lett., 1996,77, 1905

    1996

  12. [21]

    Sanejouand, inElastic Network Models: Theoretical and Empirical Foundations, Hu- mana Press, Totowa, N.J., 2013, pp

    Y.-H. Sanejouand, inElastic Network Models: Theoretical and Empirical Foundations, Hu- mana Press, Totowa, N.J., 2013, pp. 601–616

    2013

  13. [22]

    Togashi and H

    Y. Togashi and H. Flechsig,Int. J. Mol. Sci., 2018,19, 3899

    2018

  14. [23]

    Micheletti, P

    C. Micheletti, P. Carloni and A. Maritan,Proteins: Struct., Funct., Bioinf., 2004,55, 635– 645

    2004

  15. [24]

    Micheletti, F

    C. Micheletti, F. Cecconi, A. Flammini and A. Maritan,Protein Sci., 2002,11, 1878–1887. 33

    2002

  16. [25]

    Haliloglu, I

    T. Haliloglu, I. Bahar and B. Erman,Phys. Rev. Lett., 1997,79, 3090

    1997

  17. [26]

    Austin, K

    R. Austin, K. Beeson, L. Eisenstein, H. Frauenfelder and I. Gunsalus,Biochemistry, 1975,14, 5355–5373

    1975

  18. [27]

    A.-A. Ali, A. Gulzar, S. Wolf and G. Stock,J. Phys. Chem. Lett., 2022,13, 9862–9868

    2022

  19. [28]

    Stock and P

    G. Stock and P. Hamm,Philos. Trans. R. Soc. B: Biol. Sci., 2018,373, 20170187

    2018

  20. [29]

    Buchenberg, F

    S. Buchenberg, F. Sittel and G. Stock,Proc. Natl. Acad. Sci., 2017,114, E6804–E6811

    2017

  21. [30]

    Xing,Phys

    J. Xing,Phys. Rev. Lett., 2007,99, 168103

    2007

  22. [31]

    Hathcock, Q

    D. Hathcock, Q. Yu, B. A. Mello, D. Amin, G. Hazelbauer and Y. Tu,Proc. Natl Acad. Sci. USA, 2023,120, e2303115120

    2023

  23. [32]

    Goodey and S

    N. Goodey and S. Benkovic,Nat. Chem. Biol., 2008,4, 474–482

    2008

  24. [33]

    Motlagh, J

    H. Motlagh, J. Wrabl, J. Li and V. Hilser,Nature, 2014,508, 331–339

    2014

  25. [34]

    Cecconi, P

    F. Cecconi, P. De Los Rios and F. Piazza,J. Phys. Chem. B, 2007,111, 11057–11063

    2007

  26. [35]

    Barducci and P

    A. Barducci and P. De Los Rios,Curr. Opin. Struct. Biol., 2015,30, 161–169

    2015

  27. [36]

    Stevens and Y

    A. Stevens and Y. He,Int. J. Mol. Sci., 2022,23, 1454

    2022

  28. [37]

    Lockless and R

    S. Lockless and R. Ranganathan,Science, 1999,286, 295–299

    1999

  29. [38]

    Ashkinadze, H

    D. Ashkinadze, H. Kadavath, A. Pokharna, C. Chi, M. Friedmann, D. Strotz, P. Kumari, M. Minges, R. Cadalbert, S. K¨ oniglet al.,Nat. Commum., 2022,13, 6232

    2022

  30. [39]

    Amacher, P

    J. Amacher, P. Cushing, C. Bahl, T. Beck and D. Madden,J. Biol. Chem., 2013,288, 5114– 5126

    2013

  31. [40]

    Kozlov, D

    G. Kozlov, D. Banville, K. Gehring and I. Ekiel,J. Mol. Biol., 2002,320, 813–820

    2002

  32. [41]

    Ivarsson,FEBS Lett., 2012,586, 2638–2647

    Y. Ivarsson,FEBS Lett., 2012,586, 2638–2647

    2012

  33. [42]

    Bahar, A

    I. Bahar, A. R. Atilgan and B. Erman,Folding & Design, 1997,2, 173–181

    1997

  34. [43]

    Bechinger, R

    C. Bechinger, R. Di Leonardo, H. L¨ owen, C. Reichhardt, G. Volpe and G. Volpe,Rev. Mod. Phys., 2016,88, 045006

    2016

  35. [44]

    Digregorio, D

    P. Digregorio, D. Levis, A. Suma, L. Cugliandolo, G. Gonnella and I. Pagonabarraga,Phys. Rev. Lett., 2018,121, 098003

    2018

  36. [45]

    Caprini, U

    L. Caprini, U. Marini Bettolo Marconi and A. Puglisi,Phys. Rev. Lett., 2020,124, 078001

    2020

  37. [46]

    Szamel,Phys

    G. Szamel,Phys. Rev. E, 2014,90, 012111

    2014

  38. [47]

    Maggi, M

    C. Maggi, M. Paoluzzi, N. Pellicciotta, A. Lepore, L. Angelani and R. Di Leonardo,Phys. Rev. Lett., 2014,113, 238303

    2014

  39. [48]

    Caprini and U

    L. Caprini and U. Marconi Marini Bettolo,Soft Matter, 2018,14, 9044–9054. 34

    2018

  40. [49]

    Bahar, A

    I. Bahar, A. R. Atilgan and B. Erman,Folding and Design, 1997,2, 173–181

    1997

  41. [50]

    Burioni, D

    R. Burioni, D. Cassi, F. Cecconi and A. Vulpiani,Proteins: Struct., Func., Bioinf., 2004,55, 529–535

    2004

  42. [51]

    Kundu, J

    S. Kundu, J. S. Melton, D. C. Sorensen and G. N. Phillips Jr,Biophys. J., 2002,83, 723–732

    2002

  43. [52]

    Cecconi, G

    F. Cecconi, G. Costantini, C. Guardiani, M. Baldovin and A. Vulpiani,Phys. Biol., 2023,20, 056002

    2023

  44. [53]

    C. Yuan, H. Chen and D. Kihara,BMC Bioinform., 2012,13, 1–13

    2012

  45. [54]

    Mandal, B

    S. Mandal, B. Liebchen and H. L¨ owen,Phys. Rev. Lett., 2019,123, 228001

    2019

  46. [55]

    Marconi Marini Bettolo and C

    U. Marconi Marini Bettolo and C. Maggi,Soft Matter, 2015,11, 8768–8781

    2015

  47. [56]

    Caprini and U

    L. Caprini and U. Marconi Marini Bettolo,Phys. Rev. R, 2020,2, 033518

    2020

  48. [57]

    Caprini, U

    L. Caprini, U. Marini Bettolo Marconi, A. Puglisi and H. L¨ owen,J. Chem. Phys., 2023,159, year

    2023

  49. [58]

    Ichiye and M

    T. Ichiye and M. Karplus,Proteins: Struc, Func., Bioinf., 1991,11, 205–217

    1991

  50. [59]

    Bozovic, C

    O. Bozovic, C. Zanobini, A. Gulzar, B. Jankovic, D. Buhrke, M. Post, S. Wolf, G. Stock and P. Hamm,Proc. Natl. Acad. Sci. USA, 2020,117, 26031–26039

    2020

  51. [60]

    Buchli, S

    B. Buchli, S. Waldauer, R. Walser, M. L. Donten, R. Pfister, N. Bl¨ ochliger, S. Steiner, A. Caflisch, O. Zerbe and P. Hamm,Proc. Natl. Acad. Sci. USA, 2013,110, 11725–11730

    2013

  52. [61]

    Buchenberg, V

    S. Buchenberg, V. Knecht, R. Walser, P. Hamm and G. Stock,J. Physi. Chem. B, 2014,118, 13468–13476

    2014

  53. [62]

    Pearl,Causality, Cambridge university press, 2009

    J. Pearl,Causality, Cambridge university press, 2009

    2009

  54. [63]

    Marini Bettolo Marconi, A

    U. Marini Bettolo Marconi, A. Puglisi, L. Rondoni and A. Vulpiani,Phys. Rep., 2008,461, 111–195

    2008

  55. [64]

    Baldovin, F

    M. Baldovin, F. Cecconi and A. Vulpiani,Phys. Rev. Res., 2020,2, 043436

    2020

  56. [65]

    Schreiber,Phys

    T. Schreiber,Phys. Rev. Lett., 2000,85, 461

    2000

  57. [66]

    Paluˇ s, V

    M. Paluˇ s, V. Kom´ arek, Z. Hrnˇ c ´ ıˇ r and K.ˇStˇ erbov´ a,Phys. Rev. E, 2001,63, 046211

    2001

  58. [67]

    Hacisuleyman and B

    A. Hacisuleyman and B. Erman,Proteins: Struct. Func. Bioinf., 2017,85, 1056–1064

    2017

  59. [68]

    Hacisuleyman and B

    A. Hacisuleyman and B. Erman,PLoS Comput. Biol., 2017,13, e1005319

    2017

  60. [69]

    J. Sun, D. Taylor and E. M. Bollt,SIAM J. Appl. Dyn., 2015,14, 73–106

    2015

  61. [70]

    Caprini, U

    L. Caprini, U. Marconi Marini Bettolo, A. Puglisi and A. Vulpiani,J. Stat. Mech.: Theory and Exp., 2019,2019, 053203

    2019

  62. [71]

    Shankar and M

    S. Shankar and M. Marchetti,Phys. Rev. E, 2018,98, 020604. 35

    2018

  63. [72]

    Dabelow, S

    L. Dabelow, S. Bo and R. Eichhorn,Phys. Rev. X, 2019,9, 021009

    2019

  64. [73]

    Mandal, K

    D. Mandal, K. Klymko and M. DeWeese,Phys. Rev. Lett., 2017,119, 258001

    2017

  65. [74]

    Caprini, U

    L. Caprini, U. M. B. M. Marconi, A. Puglisi and A. Vulpiani,Phys. Rev. Lett., 2018,121, 139801. 36

    2018

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.