pith. sign in

arxiv: 2606.04452 · v1 · pith:VXAODMNKnew · submitted 2026-06-03 · ⚛️ physics.chem-ph

DeltaDiff: Training-Free, Physics-Guided Machine Learning for Predicting Mutant Protein Structures

Yajie Cai , Yanbin Wang , Ming Chen This is my paper

Pith reviewed 2026-06-28 04:13 UTC · model grok-4.3

classification ⚛️ physics.chem-ph
keywords mutant protein structuresdiffusion modelsphysics-guided inferencetraining-free predictionprotein conformation changesmutation effectsmachine learning for chemistry
0
0 comments X

The pith

DeltaDiff adds mutation-aware physical guidance to a pre-trained diffusion model to predict mutant protein structures without retraining.

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

The paper presents DeltaDiff as a way to generate structures for proteins that differ from a known wild-type by only one or a few mutations. Existing diffusion models for protein structure often fail on these cases because the sequences are too similar yet the structural shifts can be large and nonlocal. DeltaDiff injects physical rules that encode the mutation's expected effects directly into the inference process of an unchanged baseline model. Tests on three systems with documented conformational rearrangements show that the guided predictions recover the key changes. This removes the need to collect new training data or retrain the network for each mutant.

Core claim

DeltaDiff is a physics-guided inference framework that incorporates mutation-aware physical guidance into a baseline diffusion model. Evaluated on Chignolin T8P, Novispirin G-10, and BBL D162N, systems that each exhibit nonlocal structural changes, DeltaDiff captures the mutation-induced conformational shifts without any retraining or fine-tuning of the baseline model.

What carries the argument

DeltaDiff, the inference-time framework that augments a baseline diffusion model with mutation-aware physical guidance.

If this is right

  • Mutant structure prediction becomes feasible at a fraction of the cost of conventional modeling or experimental methods.
  • Rational design of mutants can proceed by generating candidate structures directly from sequence changes.
  • The same guidance approach may apply to other single-site or few-site sequence variants that induce conformational shifts.
  • Training data collection and model retraining are no longer required for each new mutant of interest.

Where Pith is reading between the lines

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

  • Physical priors supplied at inference time can substitute for task-specific training data in structure-prediction networks.
  • The method may generalize to other biomolecules where small sequence edits produce large structural effects.
  • Integration with additional pre-trained models could further lower the barrier to exploring mutational landscapes.
  • The approach highlights a route to make existing large models more adaptable without additional compute for retraining.

Load-bearing premise

A pre-trained diffusion model can be steered at inference time by physical rules that encode mutation effects and thereby recover nonlocal structural rearrangements that the model was never explicitly trained to produce.

What would settle it

High-resolution experimental structures of the three tested mutants that deviate substantially from the DeltaDiff predictions in the regions reported to change conformation.

Figures

Figures reproduced from arXiv: 2606.04452 by Ming Chen, Yajie Cai, Yanbin Wang.

Figure 1
Figure 1. Figure 1: Schematic illustration of mutant-structure generation with physically guided diffu [PITH_FULL_IMAGE:figures/full_fig_p008_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) Folded state of chignolin structure (PDB ID: 1UAO) and two representative T8P mutant structures generated by DeltaDiff. Asp3, Gly7, and Thr8/Pro8 are labeled, with Thr8/Pro8 indicating the mutation site. The two mutant structures are overlaid with their closest MD-sampled structures in the background. (b) Rosetta relaxed T8P structures are projected onto Rosetta score and Cα RMSD to the wild type PDB s… view at source ↗
Figure 3
Figure 3. Figure 3: (a) Relaxed I10G structures are projected onto the two-dimensional feature space defined by Rosetta score and Cα RMSD to the mutant 1HU6 reference structure. The dashed line marks the median DeltaDiff relaxed-score cutoff used for screening. The inset shows a representative DeltaDiff-generated structure with the lowest RMSD to 1HU6 and a low Rosetta score (orange), overlaid with the 1HU6 structure (green).… view at source ↗
Figure 4
Figure 4. Figure 4: (a) Conformational change of the BBL D162N. The wild-type structure (PDB ID: 2WXC) is shown in red, the dominant D162N MD structure in green, and a represen￾tative DeltaDiff-generated D162N structure in orange. Key residues Ala130 and Arg157 are highlighted in blue, Gly153 is highlighted in yellow, and the mutation site Asp162/Asn162 is highlighted in purple. (b) Relaxed D162N structures projected onto the… view at source ↗
read the original abstract

Determining mutant protein structures is critical for understanding the mechanistic roles of mutations in biochemical processes. However, experimental characterization and conventional theoretical modeling are often expensive and time-consuming. Recent advances in machine learning provide new opportunities to efficiently predict protein structures from primary sequences. Nevertheless, applying these models to proteins with single-site or few-site mutations remains challenging because mutant sequences are often highly similar to their wild-type counterparts. Here, we introduce DeltaDiff, a physics-guided inference framework for mutant-structure generation that incorporates mutation-aware physical guidance into a baseline diffusion model. We evaluate DeltaDiff on three representative systems: Chignolin T8P, Novispirin G-10, and BBL D162N. All three examples involve nonlocal structural changes, making accurate mutant-structure prediction challenging. DeltaDiff captures key mutation-induced conformational changes without requiring retraining or fine-tuning of the baseline model. These results establish a foundation for efficient mutant-structure prediction at a fraction of the cost of conventional methods, facilitating rational mutant design.

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

2 major / 2 minor

Summary. The manuscript introduces DeltaDiff, a training-free physics-guided inference framework that augments a baseline diffusion model with mutation-aware physical guidance during sampling to predict structures of mutant proteins. It is evaluated on three systems exhibiting nonlocal conformational changes (Chignolin T8P, Novispirin G-10, BBL D162N) and claims that the guidance enables capture of key mutation-induced changes without retraining or fine-tuning the baseline model, at lower cost than conventional approaches.

Significance. If the central claim holds with quantitative support, the training-free aspect would be a notable strength for efficient mutant-structure prediction in computational biophysics and protein design, avoiding the expense of retraining large diffusion models while leveraging existing ones. The focus on nonlocal changes addresses a known challenge in sequence-similar mutants.

major comments (2)
  1. [Abstract / Results] The provided description (including abstract) contains no quantitative metrics (e.g., RMSD, GDT-TS, or contact-map accuracy) or direct comparisons to the unaugmented baseline diffusion model or other mutant-prediction methods. This is load-bearing for the claim that the guidance accurately captures nonlocal changes on the three systems.
  2. [Methods] The implementation of 'mutation-aware physical guidance' is not specified with equations or pseudocode (e.g., how mutation effects enter the score function or energy terms during inference). Without this, it is not possible to verify that the approach is truly training-free and non-circular for nonlocal effects.
minor comments (2)
  1. Clarify the baseline diffusion model used (architecture, training data) and any hyper-parameters of the guidance term.
  2. Add a table or figure summarizing structural metrics for wild-type vs. mutant predictions across the three systems.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive report and the recommendation for major revision. We agree that quantitative metrics and explicit methodological details are necessary to support the central claims. Below we respond point-by-point and indicate the revisions that will be made.

read point-by-point responses

  1. Referee: [Abstract / Results] The provided description (including abstract) contains no quantitative metrics (e.g., RMSD, GDT-TS, or contact-map accuracy) or direct comparisons to the unaugmented baseline diffusion model or other mutant-prediction methods. This is load-bearing for the claim that the guidance accurately captures nonlocal changes on the three systems.

    Authors: We agree that quantitative support is essential. The current manuscript reports only qualitative observations of conformational changes on the three systems. In the revised version we will add RMSD, GDT-TS, and contact-map accuracy metrics for both DeltaDiff and the unaugmented baseline diffusion model on all three test cases (Chignolin T8P, Novispirin G-10, BBL D162N). We will also include a direct comparison table and, where feasible, reference to existing mutant-prediction methods. These additions will be placed in a new Results subsection and summarized in the abstract. revision: yes

  2. Referee: [Methods] The implementation of 'mutation-aware physical guidance' is not specified with equations or pseudocode (e.g., how mutation effects enter the score function or energy terms during inference). Without this, it is not possible to verify that the approach is truly training-free and non-circular for nonlocal effects.

    Authors: We acknowledge that the Methods section currently describes the guidance at a high level without explicit equations. In the revision we will insert a dedicated subsection containing the precise formulation: the modified score function s_θ(x_t, t; ΔE_mut) where the mutation-induced energy term ΔE_mut is added to the denoising step, together with pseudocode for the guided sampling loop. This will make clear that no model parameters are updated and that the guidance remains non-circular because it uses only the fixed baseline model plus a physics-based energy evaluator. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents DeltaDiff as a training-free inference framework that augments an existing baseline diffusion model with mutation-aware physical guidance. No equations, derivations, fitted parameters, or self-citation chains are described in the provided abstract or claim summary. The central claim reduces to an empirical demonstration on three systems rather than any mathematical reduction of a prediction to its own inputs. The method is therefore self-contained against external benchmarks with no load-bearing circular steps.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Since only the abstract is available, no free parameters, axioms, or invented entities can be identified from the provided information.

pith-pipeline@v0.9.1-grok · 5734 in / 1087 out tokens · 70578 ms · 2026-06-28T04:13:59.028504+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

158 extracted references · 110 canonical work pages

  1. [1]

    Plos one , volume=

    Using AlphaFold to predict the impact of single mutations on protein stability and function , author=. Plos one , volume=. 2023 , publisher=

    2023

  2. [2]

    Journal of Chemical Information and Modeling , volume=

    Evaluation of AlphaFold 3’s protein--protein complexes for predicting binding free energy changes upon mutation , author=. Journal of Chemical Information and Modeling , volume=. 2024 , publisher=

    2024

  3. [3]

    Computers in biology and medicine , volume=

    Integration of persistent Laplacian and pre-trained transformer for protein solubility changes upon mutation , author=. Computers in biology and medicine , volume=. 2024 , publisher=

    2024

  4. [4]

    Nature communications , volume=

    Machine learning coarse-grained potentials of protein thermodynamics , author=. Nature communications , volume=. 2023 , publisher=

    2023

  5. [5]

    Proceedings of the National Academy of Sciences , volume=

    Protein language models trained on biophysical dynamics inform mutation effects , author=. Proceedings of the National Academy of Sciences , volume=. 2026 , publisher=

    2026

  6. [6]

    Science , volume=

    Evolutionary-scale prediction of atomic-level protein structure with a language model , author=. Science , volume=. 2023 , publisher=

    2023

  7. [7]

    Science , volume=

    Scalable emulation of protein equilibrium ensembles with generative deep learning , author=. Science , volume=. 2025 , publisher=

    2025

  8. [8]

    Nature , volume=

    De novo design of protein structure and function with RFdiffusion , author=. Nature , volume=. 2023 , publisher=

    2023

  9. [9]

    International Conference on Learning Representations , volume=

    Str2str: A score-based framework for zero-shot protein conformation sampling , author=. International Conference on Learning Representations , volume=

  10. [10]

    bioRxiv , pages=

    Rapid sequence-based screening of structure-disrupting protein mutations , author=. bioRxiv , pages=. 2026 , publisher=

    2026

  11. [11]

    Structure , volume=

    10 residue folded peptide designed by segment statistics , author=. Structure , volume=. 2004 , publisher=

    2004

  12. [12]

    RSC advances , volume=

    Mutation-induced change in chignolin stability from -turn to -turn , author=. RSC advances , volume=. 2020 , publisher=

    2020

  13. [13]

    Nature Methods , volume =

    Deep mutational scanning: a new style of protein science , author =. Nature Methods , volume =. 2014 , publisher =. doi:10.1038/nmeth.3027 , url =

    work page doi:10.1038/nmeth.3027 2014

  14. [14]

    Nature Reviews Molecular Cell Biology , volume =

    Exploring protein fitness landscapes by directed evolution , author =. Nature Reviews Molecular Cell Biology , volume =. 2009 , publisher =. doi:10.1038/nrm2805 , url =

    work page doi:10.1038/nrm2805 2009

  15. [15]

    Current Opinion in Structural Biology , volume =

    Stability effects of mutations and protein evolvability , author =. Current Opinion in Structural Biology , volume =. 2009 , publisher =. doi:10.1016/j.sbi.2009.08.003 , url =

    work page doi:10.1016/j.sbi.2009.08.003 2009

  16. [16]

    Current Opinion in Structural Biology , volume =

    Allostery without a conformational change? Revisiting the paradigm , author =. Current Opinion in Structural Biology , volume =. 2015 , publisher =. doi:10.1016/j.sbi.2014.11.005 , url =

    work page doi:10.1016/j.sbi.2014.11.005 2015

  17. [17]

    Annual Review of Biophysics , volume =

    The Role of Conformational Dynamics and Allostery in Modulating Protein Evolution , author =. Annual Review of Biophysics , volume =. 2020 , publisher =. doi:10.1146/annurev-biophys-052118-115517 , url =

    work page doi:10.1146/annurev-biophys-052118-115517 2020

  18. [18]

    Highly Accurate Protein Structure Prediction with

    Highly accurate protein structure prediction with AlphaFold , author =. Nature , volume =. 2021 , publisher =. doi:10.1038/s41586-021-03819-2 , url =

    work page doi:10.1038/s41586-021-03819-2 2021

  19. [19]

    Science , volume =

    Accurate prediction of protein structures and interactions using a three-track neural network , author =. Science , volume =. 2021 , publisher =. doi:10.1126/science.abj8754 , url =

    work page doi:10.1126/science.abj8754 2021

  20. [20]

    Nature , volume =

    Accurate structure prediction of biomolecular interactions with AlphaFold 3 , author =. Nature , volume =. 2024 , publisher =. doi:10.1038/s41586-024-07487-w , url =

    work page doi:10.1038/s41586-024-07487-w 2024

  21. [21]

    2020 , eprint =

    Score-Based Generative Modeling through Stochastic Differential Equations , author =. 2020 , eprint =

    2020

  22. [22]

    Nature Computational Science , volume =

    Score-based generative modeling for de novo protein design , author =. Nature Computational Science , volume =. 2023 , publisher =. doi:10.1038/s43588-023-00440-3 , url =

    work page doi:10.1038/s43588-023-00440-3 2023

  23. [23]

    Lawrence Zitnick, Jerry Ma, and Rob Fergus

    Biological structure and function emerge from scaling unsupervised learning to 250 million protein sequences , author =. Proceedings of the National Academy of Sciences , volume =. 2021 , publisher =. doi:10.1073/pnas.2016239118 , url =

    work page doi:10.1073/pnas.2016239118 2021

  24. [24]

    2021 , publisher =

    Language models enable zero-shot prediction of the effects of mutations on protein function , author =. 2021 , publisher =. doi:10.1101/2021.07.09.450648 , url =

    work page doi:10.1101/2021.07.09.450648 2021

  25. [25]

    2021 , publisher =

    MSA Transformer , author =. 2021 , publisher =. doi:10.1101/2021.02.12.430858 , url =

    work page doi:10.1101/2021.02.12.430858 2021

  26. [26]

    2022 , eprint =

    Tranception: protein fitness prediction with autoregressive transformers and inference-time retrieval , author =. 2022 , eprint =

    2022

  27. [27]

    Nature Genetics , volume =

    Genome-wide prediction of disease variant effects with a deep protein language model , author =. Nature Genetics , volume =. 2023 , publisher =. doi:10.1038/s41588-023-01465-0 , url =

    work page doi:10.1038/s41588-023-01465-0 2023

  28. [28]

    Nature Structural &; Molecular Biology , volume =

    Can AlphaFold2 predict the impact of missense mutations on structure? , author =. Nature Structural &; Molecular Biology , volume =. 2022 , publisher =. doi:10.1038/s41594-021-00714-2 , url =

    work page doi:10.1038/s41594-021-00714-2 2022

  29. [29]

    Nature Chemical Biology , volume =

    The power and pitfalls of AlphaFold2 for structure prediction beyond rigid globular proteins , author =. Nature Chemical Biology , volume =. 2024 , publisher =. doi:10.1038/s41589-024-01638-w , url =

    work page doi:10.1038/s41589-024-01638-w 2024

  30. [30]

    Physical Review Letters , volume =

    AlphaFold2 Can Predict Single-Mutation Effects , author =. Physical Review Letters , volume =. 2023 , publisher =. doi:10.1103/PhysRevLett.131.218401 , url =

    work page doi:10.1103/physrevlett.131.218401 2023

  31. [31]

    2021 , eprint=

    Diffusion Models Beat GANs on Image Synthesis , author=. 2021 , eprint=

    2021

  32. [32]

    Chemical Reviews , volume =

    Machine Learning Force Fields , author =. Chemical Reviews , volume =. 2021 , publisher =. doi:10.1021/acs.chemrev.0c01111 , url =

    work page doi:10.1021/acs.chemrev.0c01111 2021

  33. [33]

    Nat Commun , volume =

    E(3)-equivariant graph neural networks for data-efficient and accurate interatomic potentials , author =. Nature Communications , volume =. 2022 , publisher =. doi:10.1038/s41467-022-29939-5 , url =

    work page doi:10.1038/s41467-022-29939-5 2022

  34. [34]

    2022 , eprint=

    TorchMD-NET: Equivariant Transformers for Neural Network based Molecular Potentials , author=. 2022 , eprint=

    2022

  35. [35]

    Nature Communications , volume=

    Engineering protein-based therapeutics through structural and chemical design , author=. Nature Communications , volume=. 2023 , doi=

    2023

  36. [36]

    Protein engineering , volume=

    Impact of single-residue mutations on the structure and function of ovispirin/novispirin antimicrobial peptides , author=. Protein engineering , volume=. 2002 , publisher=

    2002

  37. [37]

    Journal of molecular biology , volume=

    The folding mechanism of BBL: Plasticity of transition-state structure observed within an ultrafast folding protein family , author=. Journal of molecular biology , volume=. 2009 , publisher=

    2009

  38. [38]

    Science , volume =

    Filamentous Fusion Phage: Novel Expression Vectors That Display Cloned Antigens on the Virion Surface , author =. Science , volume =. 1985 , publisher =. doi:10.1126/science.4001944 , url =

    work page doi:10.1126/science.4001944 1985

  39. [39]

    Proceedings of the National Academy of Sciences , volume =

    Using deep sequencing to characterize the biophysical mechanism of a transcriptional regulatory sequence , author =. Proceedings of the National Academy of Sciences , volume =. 2010 , publisher =. doi:10.1073/pnas.1004290107 , url =

    work page doi:10.1073/pnas.1004290107 2010

  40. [40]

    Nature Structural Biology , volume =

    The way to NMR structures of proteins , author =. Nature Structural Biology , volume =. 2001 , publisher =. doi:10.1038/nsb1101-923 , url =

    work page doi:10.1038/nsb1101-923 2001

  41. [41]

    Science , volume =

    The Resolution Revolution , author =. Science , volume =. 2014 , publisher =. doi:10.1126/science.1251652 , url =

    work page doi:10.1126/science.1251652 2014

  42. [42]

    Crystallography Made Crystal Clear: A Guide for Users of Macromolecular Models , author=

  43. [43]

    Nature Protocols , volume =

    Using circular dichroism spectra to estimate protein secondary structure , author =. Nature Protocols , volume =. 2006 , publisher =. doi:10.1038/nprot.2006.202 , url =

    work page doi:10.1038/nprot.2006.202 2006

  44. [44]

    Annual Review of Biophysics , volume =

    Biomolecular Simulation: A Computational Microscope for Molecular Biology , author =. Annual Review of Biophysics , volume =. 2012 , publisher =. doi:10.1146/annurev-biophys-042910-155245 , url =

    work page doi:10.1146/annurev-biophys-042910-155245 2012

  45. [45]

    Accounts of Chemical Research , volume =

    Molecular Dynamics Simulations of Biomolecules , author =. Accounts of Chemical Research , volume =. 2002 , publisher =. doi:10.1021/ar020082r , url =

    work page doi:10.1021/ar020082r 2002

  46. [46]

    2021 , publisher =

    Protein complex prediction with AlphaFold-Multimer , author =. 2021 , publisher =. doi:10.1101/2021.10.04.463034 , url =

    work page doi:10.1101/2021.10.04.463034 2021

  47. [47]

    Biopolymers , volume =

    Dictionary of protein secondary structure: Pattern recognition of hydrogen‐bonded and geometrical features , author =. Biopolymers , volume =. 1983 , publisher =. doi:10.1002/bip.360221211 , url =

    work page doi:10.1002/bip.360221211 1983

  48. [48]

    PRX Life , volume =

    ExEnDiff: An Experiment-Guided Diffusion Model for Protein Conformational Ensemble Generation , author =. PRX Life , volume =. 2025 , publisher =. doi:10.1103/PRXLife.3.023013 , url =

    work page doi:10.1103/prxlife.3.023013 2025

  49. [49]

    The Journal of Physical Chemistry Letters , volume =

    Extrapolating Foundation Generative Models with Physics: A Case Study of Exploring Peptide Conformations under Protein–Environment Interactions , author =. The Journal of Physical Chemistry Letters , volume =. 2025 , publisher =. doi:10.1021/acs.jpclett.5c02567 , url =

    work page doi:10.1021/acs.jpclett.5c02567 2025

  50. [50]

    bioRxiv , year=

    ESMAdam: A Plug-and-Play All-Purpose Protein Ensemble Generator , author=. bioRxiv , year=

  51. [51]

    Nature Reviews Microbiology , volume =

    Molecular mechanisms of antibiotic resistance , author =. Nature Reviews Microbiology , volume =. 2014 , publisher =. doi:10.1038/nrmicro3380 , url =

    work page doi:10.1038/nrmicro3380 2014

  52. [52]

    Cell , volume =

    Widespread Macromolecular Interaction Perturbations in Human Genetic Disorders , author =. Cell , volume =. 2015 , publisher =. doi:10.1016/j.cell.2015.04.013 , url =

    work page doi:10.1016/j.cell.2015.04.013 2015

  53. [53]

    Nature Reviews Chemistry , volume =

    Loop dynamics and the evolution of enzyme activity , author =. Nature Reviews Chemistry , volume =. 2023 , publisher =. doi:10.1038/s41570-023-00495-w , url =

    work page doi:10.1038/s41570-023-00495-w 2023

  54. [54]

    Proceedings of the National Academy of Sciences , volume =

    Interconversion between two unrelated protein folds in the lymphotactin native state , author =. Proceedings of the National Academy of Sciences , volume =. 2008 , publisher =. doi:10.1073/pnas.0709518105 , url =

    work page doi:10.1073/pnas.0709518105 2008

  55. [55]

    Proceedings of the National Academy of Sciences , volume =

    Reversible switching between two common protein folds in a designed system using only temperature , author =. Proceedings of the National Academy of Sciences , volume =. 2023 , publisher =. doi:10.1073/pnas.2215418120 , url =

    work page doi:10.1073/pnas.2215418120 2023

  56. [56]

    Nature Reviews Microbiology , volume =

    Molecular mechanisms of antibiotic resistance revisited , author =. Nature Reviews Microbiology , volume =. 2022 , publisher =. doi:10.1038/s41579-022-00820-y , url =

    work page doi:10.1038/s41579-022-00820-y 2022

  57. [57]

    Nature Reviews Molecular Cell Biology , volume =

    Structural and functional constraints in the evolution of protein families , author =. Nature Reviews Molecular Cell Biology , volume =. 2009 , publisher =. doi:10.1038/nrm2762 , url =

    work page doi:10.1038/nrm2762 2009

  58. [58]

    Nature Methods , volume =

    High-resolution mapping of protein sequence-function relationships , author =. Nature Methods , volume =. 2010 , publisher =. doi:10.1038/nmeth.1492 , url =

    work page doi:10.1038/nmeth.1492 2010

  59. [59]

    Nucleic Acids Research , volume =

    The Protein Data Bank , author =. Nucleic Acids Research , volume =. 2000 , publisher =. doi:10.1093/nar/28.1.235 , url =

    work page doi:10.1093/nar/28.1.235 2000

  60. [60]

    Nucleic Acids Research , volume =

    AlphaFold Protein Structure Database: massively expanding the structural coverage of protein-sequence space with high-accuracy models , author =. Nucleic Acids Research , volume =. 2021 , publisher =. doi:10.1093/nar/gkab1061 , url =

    work page doi:10.1093/nar/gkab1061 2021

  61. [61]

    Current Opinion in Structural Biology , volume =

    Proteins that switch folds , author =. Current Opinion in Structural Biology , volume =. 2010 , publisher =. doi:10.1016/j.sbi.2010.06.002 , url =

    work page doi:10.1016/j.sbi.2010.06.002 2010

  62. [62]

    Structure , volume =

    Mutational Tipping Points for Switching Protein Folds and Functions , author =. Structure , volume =. 2012 , publisher =. doi:10.1016/j.str.2011.11.018 , url =

    work page doi:10.1016/j.str.2011.11.018 2012

  63. [63]

    2022 , eprint =

    Classifier-Free Diffusion Guidance , author =. 2022 , eprint =

    2022

  64. [64]

    2020 , eprint =

    Denoising Diffusion Probabilistic Models , author =. 2020 , eprint =

    2020

  65. [65]

    Proceedings of the National Academy of Sciences , volume =

    An end-to-end deep learning method for protein side-chain packing and inverse folding , author =. Proceedings of the National Academy of Sciences , volume =. 2023 , publisher =. doi:10.1073/pnas.2216438120 , url =

    work page doi:10.1073/pnas.2216438120 2023

  66. [66]

    Bioinformatics , volume =

    PyRosetta: a script-based interface for implementing molecular modeling algorithms using Rosetta , author =. Bioinformatics , volume =. 2010 , publisher =. doi:10.1093/bioinformatics/btq007 , url =

    work page doi:10.1093/bioinformatics/btq007 2010

  67. [67]

    Journal of Chemical Theory and Computation , volume =

    The Rosetta All-Atom Energy Function for Macromolecular Modeling and Design , author =. Journal of Chemical Theory and Computation , volume =. 2017 , publisher =. doi:10.1021/acs.jctc.7b00125 , url =

    work page doi:10.1021/acs.jctc.7b00125 2017

  68. [68]

    FEBS Letters , volume =

    Folding free‐energy landscape of a 10‐residue mini‐protein, chignolin , author =. FEBS Letters , volume =. 2006 , publisher =. doi:10.1016/j.febslet.2006.05.015 , url =

    work page doi:10.1016/j.febslet.2006.05.015 2006

  69. [69]

    Well-Tempered Metadynamics: A Smoothly Converging and Tunable Free-Energy Method

    Well-Tempered Metadynamics: A Smoothly Converging and Tunable Free-Energy Method , author =. Physical Review Letters , volume =. 2008 , publisher =. doi:10.1103/PhysRevLett.100.020603 , url =

    work page doi:10.1103/physrevlett.100.020603 2008

  70. [70]

    2023 , eprint =

    Steered Diffusion: A Generalized Framework for Plug-and-Play Conditional Image Synthesis , author =. 2023 , eprint =

    2023

  71. [71]

    Nature Reviews Genetics , volume =

    Missense meanderings in sequence space: a biophysical view of protein evolution , author =. Nature Reviews Genetics , volume =. 2005 , publisher =. doi:10.1038/nrg1672 , url =

    work page doi:10.1038/nrg1672 2005

  72. [72]

    Nature Reviews Genetics , volume =

    Mutational effects and the evolution of new protein functions , author =. Nature Reviews Genetics , volume =. 2010 , publisher =. doi:10.1038/nrg2808 , url =

    work page doi:10.1038/nrg2808 2010

  73. [73]

    Nature , volume =

    Mega-scale experimental analysis of protein folding stability in biology and design , author =. Nature , volume =. 2023 , publisher =. doi:10.1038/s41586-023-06328-6 , url =

    work page doi:10.1038/s41586-023-06328-6 2023

  74. [74]

    Science , volume =

    Accurate proteome-wide missense variant effect prediction with AlphaMissense , author =. Science , volume =. 2023 , publisher =. doi:10.1126/science.adg7492 , url =

    work page doi:10.1126/science.adg7492 2023

  75. [75]

    Nature , volume =

    Disease variant prediction with deep generative models of evolutionary data , author =. Nature , volume =. 2021 , publisher =. doi:10.1038/s41586-021-04043-8 , url =

    work page doi:10.1038/s41586-021-04043-8 2021

  76. [76]

    Nature Methods , volume =

    Deep generative models of genetic variation capture the effects of mutations , author =. Nature Methods , volume =. 2018 , publisher =. doi:10.1038/s41592-018-0138-4 , url =

    work page doi:10.1038/s41592-018-0138-4 2018

  77. [77]

    Nature Biotechnology , volume =

    Mutation effects predicted from sequence co-variation , author =. Nature Biotechnology , volume =. 2017 , publisher =. doi:10.1038/nbt.3769 , url =

    work page doi:10.1038/nbt.3769 2017

  78. [78]

    Nature , volume =

    Dynamic personalities of proteins , author =. Nature , volume =. 2007 , publisher =. doi:10.1038/nature06522 , url =

    work page doi:10.1038/nature06522 2007

  79. [79]

    Science , volume =

    The Energy Landscapes and Motions of Proteins , author =. Science , volume =. 1991 , publisher =. doi:10.1126/science.1749933 , url =

    work page doi:10.1126/science.1749933 1991

  80. [80]

    Science , volume =

    The Protein-Folding Problem, 50 Years On , author =. Science , volume =. 2012 , publisher =. doi:10.1126/science.1219021 , url =

    work page doi:10.1126/science.1219021 2012

Showing first 80 references.