pith. sign in

arxiv: 2509.02060 · v4 · submitted 2025-09-02 · 🧬 q-bio.BM · cs.LG

Morphology-Aware Peptide Discovery via Masked Conditional Generative Modeling

Nuno Costa , Julija Zavadlav This is my paper

Pith reviewed 2026-05-18 20:23 UTC · model grok-4.3

classification 🧬 q-bio.BM cs.LG
keywords peptide self-assemblygenerative modelingconditional variational autoencodermorphology proxiescoarse-grained molecular dynamicssequence designbiomaterial discovery
0
0 comments X

The pith

Conditioning a masked variational autoencoder on peptide descriptors generates sequences that self-assemble into targeted fibrillar or spherical shapes.

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

Peptide self-assembly offers a route to biocompatible materials, yet the sequence space is too large for direct screening of desired aggregate shapes. The work compiles a dataset of geometric and physicochemical descriptors from existing aggregation data and trains a Transformer-based conditional variational autoencoder equipped with a masking mechanism. This model generates new peptide sequences conditioned on chosen descriptor values that act as morphology proxies. After design filters and coarse-grained molecular dynamics validation, the pipeline reports an 83 percent success rate in producing peptides that match the intended fibrillar or spherical morphology.

Core claim

PepMorph trains a Transformer-based Conditional Variational Autoencoder with masking on peptide descriptors to generate novel sequences under arbitrary conditioning, then filters and validates the outputs with CG-MD simulations to achieve an 83 percent success rate for steering self-assembly toward fibrillar or spherical morphologies.

What carries the argument

The masked conditional variational autoencoder that accepts geometric and physicochemical descriptors as conditioning inputs to produce novel peptide sequences.

If this is right

  • Peptide sequences can be generated on demand for applications that require specific aggregate shapes such as fibers for scaffolds or spheres for encapsulation.
  • The method reduces reliance on brute-force enumeration of sequences by learning to map descriptors directly to assembly outcomes.
  • High validation rates under the reported protocol indicate that the chosen proxies capture enough information to guide morphology without full atomistic detail.
  • The pipeline can be reused for other morphology targets by simply changing the conditioning descriptors.

Where Pith is reading between the lines

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

  • If the proxies remain stable across different simulation resolutions, the same conditioning approach could transfer to designing larger protein assemblies or hybrid nanomaterials.
  • An inverse version of the model might start from a target morphology and output the required descriptor values, closing the loop between shape and sequence.
  • Combining the generative step with experimental synthesis feedback could iteratively improve the descriptor-to-morphology mapping.

Load-bearing premise

The extracted geometric and physicochemical descriptors function as reliable independent proxies that steer the generative model toward the intended aggregate morphologies.

What would settle it

A larger set of generated peptides run through the same CG-MD protocol yields a success rate substantially below 83 percent for the conditioned morphology class.

read the original abstract

Peptide self-assembly prediction offers a powerful bottom-up strategy for designing biocompatible, low-toxicity materials for large-scale synthesis in a broad range of biomedical and energy applications. However, screening the vast sequence space for categorization of aggregate morphology remains intractable. We introduce PepMorph, an end-to-end peptide discovery pipeline that generates novel sequences that are not only prone to aggregate but whose self-assembly is steered toward fibrillar or spherical morphologies by conditioning on isolated peptide descriptors that serve as morphology proxies. To this end, we compiled a new dataset by leveraging existing aggregation propensity datasets and extracting geometric and physicochemical descriptors. This dataset is then used to train a Transformer-based Conditional Variational Autoencoder with a masking mechanism, which generates novel peptides under arbitrary conditioning. After filtering to ensure design specifications and validation of generated sequences through coarse-grained molecular dynamics (CG-MD) simulations, PepMorph yielded 83% success rate under our CG-MD validation protocol and morphology criterion for the targeted class, showcasing its promise as a framework for application-driven peptide discovery.

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 paper introduces PepMorph, an end-to-end pipeline for morphology-aware peptide discovery. It compiles a new dataset from existing aggregation propensity sources, extracts geometric and physicochemical descriptors to serve as morphology proxies, and trains a Transformer-based masked conditional variational autoencoder to generate novel peptide sequences conditioned on arbitrary descriptor values. Generated candidates are filtered for design specifications and validated via coarse-grained molecular dynamics (CG-MD) simulations, with the authors reporting an 83% success rate under their morphology criterion for targeted fibrillar or spherical assemblies.

Significance. If the central performance claim holds after addressing the noted gaps, the work would provide a useful generative framework for steering peptide self-assembly toward application-relevant morphologies without exhaustive enumeration of sequence space. The combination of descriptor-conditioned generation with CG-MD validation is a constructive step toward physically grounded peptide design for biomaterials and biomedical uses.

major comments (2)
  1. [Abstract and Results] Abstract and Results section: The central claim of an 83% success rate under the CG-MD validation protocol is presented without dataset size, number of generated/filtered sequences, model hyperparameters, or statistical error bars. This information is required to assess whether the reported rate reflects reliable steering by the conditional model rather than post-hoc selection.
  2. [Methods and Results] Methods and Results: No quantitative evidence (mutual information, regression R², or ablation removing descriptor classes) is supplied to demonstrate that the extracted geometric and physicochemical descriptors function as independent predictors of fibrillar versus spherical morphology in CG-MD trajectories, as opposed to being largely redundant with overall aggregation propensity. Without this, it remains possible that the success rate arises from filtering rather than from the masked conditional generative mechanism.
minor comments (2)
  1. [Methods] The description of the masking mechanism within the conditional VAE would benefit from an explicit equation or pseudocode block showing how conditioning vectors are incorporated during both training and sampling.
  2. [Figures] Figure captions and axis labels for any CG-MD trajectory visualizations should explicitly state the morphology criterion thresholds used to classify success.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and detailed review. We address each major comment below and commit to revisions that enhance reproducibility and strengthen the evidence for our claims.

read point-by-point responses

  1. Referee: [Abstract and Results] Abstract and Results section: The central claim of an 83% success rate under the CG-MD validation protocol is presented without dataset size, number of generated/filtered sequences, model hyperparameters, or statistical error bars. This information is required to assess whether the reported rate reflects reliable steering by the conditional model rather than post-hoc selection.

    Authors: We agree that these details are essential for evaluating the reliability of the reported success rate. In the revised manuscript, we will expand the Abstract and Results sections to report the size of the compiled training dataset, the total number of sequences generated by the model, the number retained after filtering for design specifications, the full set of Transformer VAE hyperparameters (layers, heads, latent dimension, masking ratio, training epochs, and optimizer settings), and statistical error bars on the 83% success rate obtained via repeated CG-MD runs or bootstrapping. These additions will allow readers to distinguish the contribution of the conditional generative mechanism from post-hoc filtering. revision: yes

  2. Referee: [Methods and Results] Methods and Results: No quantitative evidence (mutual information, regression R², or ablation removing descriptor classes) is supplied to demonstrate that the extracted geometric and physicochemical descriptors function as independent predictors of fibrillar versus spherical morphology in CG-MD trajectories, as opposed to being largely redundant with overall aggregation propensity. Without this, it remains possible that the success rate arises from filtering rather than from the masked conditional generative mechanism.

    Authors: We acknowledge the importance of demonstrating that the descriptors capture morphology-specific signal beyond general aggregation propensity. In the revised manuscript, we will add quantitative analyses to the Methods and Results sections: mutual information between each descriptor and morphology labels derived from the CG-MD trajectories, linear regression R² values for predicting morphology from the descriptor set, and an ablation study comparing full-descriptor conditioning against models trained on descriptor subsets. These results will be presented alongside the existing pipeline description to clarify the independent contribution of the masked conditional generative model. revision: yes

Circularity Check

0 steps flagged

No significant circularity; external CG-MD validation is independent

full rationale

The paper extracts geometric and physicochemical descriptors from existing aggregation datasets to condition a masked Transformer VAE, generates candidate sequences, applies post-generation filtering for design specs, and then evaluates success via separate coarse-grained molecular dynamics simulations that directly assess fibrillar or spherical morphology. The reported 83% success rate is therefore measured against an external simulation protocol rather than any quantity internal to the generative model or derived from the conditioning descriptors themselves. No equations, self-citations, uniqueness theorems, or fitted parameters are shown to reduce the central claim to a tautology or to the training inputs by construction. The derivation chain is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Review performed on abstract only; no explicit free parameters, axioms, or invented entities are stated in the provided text.

pith-pipeline@v0.9.0 · 5709 in / 1047 out tokens · 28677 ms · 2026-05-18T20:23:10.923099+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.

Reference graph

Works this paper leans on

47 extracted references · 47 canonical work pages · 3 internal anchors

  1. [1]

    Science (New York, N.y.) 335(6070), 813–817 (2012) https://doi.org/10.1126/science.1205962

    Aida, T., Meijer, E.W., Stupp, S.I.: Functional Supramolecular Polymers. Science (New York, N.y.) 335(6070), 813–817 (2012) https://doi.org/10.1126/science.1205962

    work page doi:10.1126/science.1205962 2012

  2. [2]

    Chemical Society Reviews 47(10), 3721–3736 (2018) https://doi.org/10.1039/ C8CS00121A

    Okesola, B.O., Mata, A.: Multicomponent self-assembly as a tool to harness new properties from peptides and proteins in material design. Chemical Society Reviews 47(10), 3721–3736 (2018) https://doi.org/10.1039/ C8CS00121A

    work page 2018

  3. [3]

    Chemical Reviews 121(22), 13869–13914 (2021) https://doi.org/10.1021/ acs.chemrev.1c00089

    Sheehan, F., Sementa, D., Jain, A., Kumar, M., Tayarani-Najjaran, M., Kroiss, D., Ulijn, R.V.: Peptide-Based Supramolecular Systems Chemistry. Chemical Reviews 121(22), 13869–13914 (2021) https://doi.org/10.1021/ acs.chemrev.1c00089

    work page 2021

  4. [4]

    Chemical Society reviews 47(10), 3406–3420 (2018) https://doi.org/10.1039/c7cs00827a

    Makam, P., Gazit, E.: Minimalistic peptide supramolecular co-assembly: Expanding the conformational space for nanotechnology. Chemical Society reviews 47(10), 3406–3420 (2018) https://doi.org/10.1039/c7cs00827a

    work page doi:10.1039/c7cs00827a 2018

  5. [5]

    Nature Reviews Chemistry 4(1), 38–53 (2020) https://doi.org/10.1038/s41570-019-0153-8 14

    Wehner, M., W¨ urthner, F.: Supramolecular polymerization through kinetic pathway control and living chain growth. Nature Reviews Chemistry 4(1), 38–53 (2020) https://doi.org/10.1038/s41570-019-0153-8 14

    work page doi:10.1038/s41570-019-0153-8 2020

  6. [6]

    Frontiers in Bioengineering and Biotechnology 8 (2020) https://doi.org/10.3389/fbioe

    Gupta, S., Singh, I., Sharma, A.K., Kumar, P.: Ultrashort Peptide Self-Assembly: Front-Runners to Transport Drug and Gene Cargos. Frontiers in Bioengineering and Biotechnology 8 (2020) https://doi.org/10.3389/fbioe. 2020.00504

    work page doi:10.3389/fbioe 2020

  7. [7]

    Theranostics 9(11), 3249–3261 (2019) https: //doi.org/10.7150/thno.31814

    Li, S., Zou, Q., Xing, R., Govindaraju, T., Fakhrullin, R., Yan, X.: Peptide-modulated self-assembly as a versatile strategy for tumor supramolecular nanotheranostics. Theranostics 9(11), 3249–3261 (2019) https: //doi.org/10.7150/thno.31814

    work page doi:10.7150/thno.31814 2019

  8. [8]

    Nature Reviews Chemistry 6(3), 165–165 (2022) https://doi.org/10.1038/ s41570-022-00367-9

    Ashworth, C.: Plastics from proteins. Nature Reviews Chemistry 6(3), 165–165 (2022) https://doi.org/10.1038/ s41570-022-00367-9

    work page 2022

  9. [9]

    Journal of Materials Chemistry C 9(37), 12462–12488 (2021) https://doi.org/10.1039/D1TC03375A

    Boddula, R., Singh, S.P.: Peptide-based novel small molecules and polymers: Unexplored optoelectronic materials. Journal of Materials Chemistry C 9(37), 12462–12488 (2021) https://doi.org/10.1039/D1TC03375A

    work page doi:10.1039/d1tc03375a 2021

  10. [10]

    Nature Communications 7(1), 13566 (2016) https://doi.org/10.1038/ncomms13566

    Nguyen, V., Zhu, R., Jenkins, K., Yang, R.: Self-assembly of diphenylalanine peptide with controlled polariza- tion for power generation. Nature Communications 7(1), 13566 (2016) https://doi.org/10.1038/ncomms13566

    work page doi:10.1038/ncomms13566 2016

  11. [11]

    Auto-Encoding Variational Bayes

    Kingma, D.P., Welling, M.: Auto-Encoding Variational Bayes. arXiv (2022). https://doi.org/10.48550/arXiv. 1312.6114

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv 2022

  12. [12]

    In: Proceedings of the 29th International Conference on Neural Information Processing Systems - Volume 2

    Sohn, K., Yan, X., Lee, H.: Learning structured output representation using deep conditional generative models. In: Proceedings of the 29th International Conference on Neural Information Processing Systems - Volume 2. NIPS’15, vol. 2, pp. 3483–3491. MIT Press, Cambridge, MA, USA (2015)

    work page 2015

  13. [13]

    PepCVAE: Semi-Supervised Targeted Design of Antimicrobial Peptide Sequences

    Das, P., Wadhawan, K., Chang, O., Sercu, T., Santos, C.D., Riemer, M., Chenthamarakshan, V., Padhi, I., Mojsilovic, A.: PepCVAE: Semi-Supervised Targeted Design of Antimicrobial Peptide Sequences. arXiv (2018). https://doi.org/10.48550/arXiv.1810.07743

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1810.07743 2018

  14. [14]

    Nature Communications 14(1), 1453 (2023) https://doi.org/10.1038/ s41467-023-36994-z

    Szymczak, P., Mo˙ zejko, M., Grzegorzek, T., Jurczak, R., Bauer, M., Neubauer, D., Sikora, K., Michal- ski, M., Sroka, J., Setny, P., Kamysz, W., Szczurek, E.: Discovering highly potent antimicrobial peptides with deep generative model HydrAMP. Nature Communications 14(1), 1453 (2023) https://doi.org/10.1038/ s41467-023-36994-z

    work page 2023

  15. [15]

    In: International Conference on Learning Representations (2018)

    Ivanov, O., Figurnov, M., Vetrov, D.: Variational Autoencoder with Arbitrary Conditioning. In: International Conference on Learning Representations (2018)

    work page 2018

  16. [16]

    Pattern Recognition 147, 110113 (2024) https://doi.org/10.1016/ j.patcog.2023.110113

    Ramchandran, S., Tikhonov, G., L¨ onnroth, O., Tiikkainen, P., L¨ ahdesm¨ aki, H.: Learning conditional varia- tional autoencoders with missing covariates. Pattern Recognition 147, 110113 (2024) https://doi.org/10.1016/ j.patcog.2023.110113

    work page arXiv 2024

  17. [17]

    Advanced Science 10(31), 2301544 (2023) https: //doi.org/10.1002/advs.202301544

    Wang, J., Liu, Z., Zhao, S., Tengyan Xu, Wang, H., Li, S.Z., Li, W.: Deep Learning Empowers the Discovery of Self-Assembling Peptides with Over 10 Trillion Sequences. Advanced Science 10(31), 2301544 (2023) https: //doi.org/10.1002/advs.202301544

    work page doi:10.1002/advs.202301544 2023

  18. [18]

    Briefings in Bioinformatics 24(6), 409 (2023) https://doi.org/10.1093/bib/ bbad409

    Liu, Z., Wang, J., Luo, Y., Zhao, S., Li, W., Li, S.Z.: Efficient prediction of peptide self-assembly through sequential and graphical encoding. Briefings in Bioinformatics 24(6), 409 (2023) https://doi.org/10.1093/bib/ bbad409

    work page doi:10.1093/bib/ 2023

  19. [19]

    doi: 10.1021/acs.jctc

    van Teijlingen, A., Tuttle, T.: Beyond Tripeptides Two-Step Active Machine Learning for Very Large Data sets. Journal of Chemical Theory and Computation 17(5), 3221–3232 (2021) https://doi.org/10.1021/acs.jctc. 1c00159

    work page doi:10.1021/acs.jctc 2021

  20. [20]

    Nature Chemistry 7(1), 30–37 (2015) https://doi.org/10.1038/nchem.2122

    Frederix, P.W.J.M., Scott, G.G., Abul-Haija, Y.M., Kalafatovic, D., Pappas, C.G., Javid, N., Hunt, N.T., Ulijn, R.V., Tuttle, T.: Exploring the sequence space for (tri-)peptide self-assembly to design and discover new hydrogels. Nature Chemistry 7(1), 30–37 (2015) https://doi.org/10.1038/nchem.2122

    work page doi:10.1038/nchem.2122 2015

  21. [21]

    Nature Machine Intelligence 6(12), 1487–1500 (2024) https://doi.org/10.1038/s42256-024-00928-1 15

    Njirjak, M., ˇZuˇ zi´ c, L., Babi´ c, M., Jankovi´ c, P., Otovi´ c, E., Kalafatovic, D., Mauˇ sa, G.: Reshaping the discovery of self-assembling peptides with generative AI guided by hybrid deep learning. Nature Machine Intelligence 6(12), 1487–1500 (2024) https://doi.org/10.1038/s42256-024-00928-1 15

    work page doi:10.1038/s42256-024-00928-1 2024

  22. [22]

    Computers in Biology and Medicine 133, 104391 (2021) https://doi.org/10.1016/j.compbiomed.2021.104391

    Mathur, D., Kaur, H., Dhall, A., Sharma, N., Raghava, G.P.S.: SAPdb: A database of short peptides and the corresponding nanostructures formed by self-assembly. Computers in Biology and Medicine 133, 104391 (2021) https://doi.org/10.1016/j.compbiomed.2021.104391

    work page doi:10.1016/j.compbiomed.2021.104391 2021

  23. [23]

    JACS Au 4(9), 3567–3580 (2024) https://doi.org/10.1021/jacsau.4c00501

    Wang, J., Liu, Z., Zhao, S., Zhang, Y., Xu, T., Li, S.Z., Li, W.: Aggregation Rules of Short Peptides. JACS Au 4(9), 3567–3580 (2024) https://doi.org/10.1021/jacsau.4c00501

    work page doi:10.1021/jacsau.4c00501 2024

  24. [24]

    Nucleic Acids Research 51(W1), 432–437 (2023) https://doi

    Rey, J., Murail, S., de Vries, S., Derreumaux, P., Tuffery, P.: PEP-FOLD4: A pH-dependent force field for peptide structure prediction in aqueous solution. Nucleic Acids Research 51(W1), 432–437 (2023) https://doi. org/10.1093/nar/gkad376

    work page doi:10.1093/nar/gkad376 2023

  25. [25]

    Protein Science 31(1), 129–140 (2022) https: //doi.org/10.1002/pro.4200

    Zardecki, C., Dutta, S., Goodsell, D.S., Lowe, R., Voigt, M., Burley, S.K.: PDB-101: Educational resources supporting molecular explorations through biology and medicine. Protein Science 31(1), 129–140 (2022) https: //doi.org/10.1002/pro.4200

    work page doi:10.1002/pro.4200 2022

  26. [26]

    Journal of Peptide Science 20(7), 453–467 (2014) https://doi.org/10.1002/psc.2633

    Dehsorkhi, A., Castelletto, V., Hamley, I.W.: Self-assembling amphiphilic peptides. Journal of Peptide Science 20(7), 453–467 (2014) https://doi.org/10.1002/psc.2633

    work page doi:10.1002/psc.2633 2014

  27. [27]

    https://zhanggroup.org/FASTA/

    Zhang Lab: FASTA Format. https://zhanggroup.org/FASTA/

    work page

  28. [28]

    Journal of Cheminformatics 10(1), 31 (2018) https://doi.org/10.1186/ s13321-018-0286-7

    Lim, J., Ryu, S., Kim, J.W., Kim, W.Y.: Molecular generative model based on conditional variational autoen- coder for de novo molecular design. Journal of Cheminformatics 10(1), 31 (2018) https://doi.org/10.1186/ s13321-018-0286-7

    work page 2018

  29. [29]

    bioRxiv (2022)

    Lin, Z., Akin, H., Rao, R., Hie, B., Zhu, Z., Lu, W., Costa, A.d.S., Fazel-Zarandi, M., Sercu, T., Candido, S., Rives, A.: Language Models of Protein Sequences at the Scale of Evolution Enable Accurate Structure Prediction. bioRxiv (2022). https://doi.org/10.1101/2022.07.20.500902

    work page doi:10.1101/2022.07.20.500902 2022

  30. [30]

    Proceedings of the National Academy of Sciences of the United States of America 98(21), 12015–12020 (2001) https://doi.org/10.1073/pnas.211536998

    Stanger, H.E., Syud, F.A., Espinosa, J.F., Giriat, I., Muir, T., Gellman, S.H.: Length-dependent stability and strand length limits in antiparallel beta -sheet secondary structure. Proceedings of the National Academy of Sciences of the United States of America 98(21), 12015–12020 (2001) https://doi.org/10.1073/pnas.211536998

    work page doi:10.1073/pnas.211536998 2001

  31. [31]

    PLoS ONE 10(8), 0134679 (2015) https://doi.org/10.1371/journal.pone.0134679

    Fam´ ılia, C., Dennison, S.R., Quintas, A., Phoenix, D.A.: Prediction of Peptide and Protein Propensity for Amyloid Formation. PLoS ONE 10(8), 0134679 (2015) https://doi.org/10.1371/journal.pone.0134679

    work page doi:10.1371/journal.pone.0134679 2015

  32. [32]

    Nature Communications 15(1), 7538 (2024) https://doi.org/10.1038/s41467-024-51933-2

    Li, T., Ren, X., Luo, X., Wang, Z., Li, Z., Luo, X., Shen, J., Li, Y., Yuan, D., Nussinov, R., Zeng, X., Shi, J., Cheng, F.: A Foundation Model Identifies Broad-Spectrum Antimicrobial Peptides against Drug-Resistant Bacterial Infection. Nature Communications 15(1), 7538 (2024) https://doi.org/10.1038/s41467-024-51933-2

    work page doi:10.1038/s41467-024-51933-2 2024

  33. [33]

    Diverse and Accurate Image Description Using a Variational Auto-Encoder with an Additive Gaussian Encoding Space

    Wang, L., Schwing, A.G., Lazebnik, S.: Diverse and Accurate Image Description Using a Variational Auto- Encoder with an Additive Gaussian Encoding Space. arXiv (2017). https://doi.org/10.48550/arXiv.1711.07068

    work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.1711.07068 2017

  34. [34]

    arXiv (2019)

    Lavda, F., Gregorov´ a, M., Kalousis, A.: Improving VAE Generations of Multimodal Data through Data- Dependent Conditional Priors. arXiv (2019). https://doi.org/10.48550/arXiv.1911.10885

    work page doi:10.48550/arxiv.1911.10885 2019

  35. [35]

    Nature communications 12(1), 6884 (2021)

    Thaler, S., Zavadlav, J.: Learning neural network potentials from experimental data via Differentiable Trajectory Reweighting. Nature communications 12(1), 6884 (2021)

    work page 2021

  36. [36]

    npj Computational Materials 10(1), 69 (2024) https://doi.org/10.1038/s41524-024-01251-4 arXiv:2308.09142 [physics]

    R¨ ocken, S., Zavadlav, J.: Accurate machine learning force fields via experimental and simulation data fusion. npj Computational Materials 10(1), 69 (2024) https://doi.org/10.1038/s41524-024-01251-4 arXiv:2308.09142 [physics]

    work page doi:10.1038/s41524-024-01251-4 2024

  37. [37]

    arXiv (2025)

    R¨ ocken, S., Burnet, A.F., Zavadlav, J.: Predicting Solvation Free Energies with an Implicit Solvent Machine Learning Potential. arXiv (2025). https://doi.org/10.48550/arXiv.2406.00183

    work page doi:10.48550/arxiv.2406.00183 2025

  38. [38]

    Journal of Chemical Theory and Computation 20(1), 411–420 (2024) https://doi.org/10.1021/acs.jctc.3c00984 16

    Coste, A., Slejko, E., Zavadlav, J., Praprotnik, M.: Developing an Implicit Solvation Machine Learning Model for Molecular Simulations of Ionic Media. Journal of Chemical Theory and Computation 20(1), 411–420 (2024) https://doi.org/10.1021/acs.jctc.3c00984 16

    work page doi:10.1021/acs.jctc.3c00984 2024

  39. [39]

    The Journal of Chemical Physics 157(24), 244103 (2022) https://doi.org/10.1063/5.0124538 arXiv:2208.10330 [physics]

    Thaler, S., Stupp, M., Zavadlav, J.: Deep Coarse-grained Potentials via Relative Entropy Minimization. The Journal of Chemical Physics 157(24), 244103 (2022) https://doi.org/10.1063/5.0124538 arXiv:2208.10330 [physics]

    work page doi:10.1063/5.0124538 2022

  40. [40]

    Chockalingam, Maneesha Aluru, and Srinivas Aluru

    Cock, P.J.A., Antao, T., Chang, J.T., Chapman, B.A., Cox, C.J., Dalke, A., Friedberg, I., Hamelryck, T., Kauff, F., Wilczynski, B., de Hoon, M.J.L.: Biopython: Freely available Python tools for computational molecular biology and bioinformatics. Bioinformatics 25(11), 1422–1423 (2009) https://doi.org/10.1093/bioinformatics/ btp163

    work page doi:10.1093/bioinformatics/ 2009

  41. [41]

    https://www.mankier.com/1/mkdssp

    Hekkelman, M.: Mkdssp: Calculate Secondary Structure for Proteins in a PDB File | Dssp Commands | Man Pages | ManKier. https://www.mankier.com/1/mkdssp

    work page

  42. [42]

    Journal of Molecular Biology 179(1), 125–142 (1984) https://doi.org/10.1016/ 0022-2836(84)90309-7

    Eisenberg, D., Schwarz, E., Komaromy, M., Wall, R.: Analysis of membrane and surface protein sequences with the hydrophobic moment plot. Journal of Molecular Biology 179(1), 125–142 (1984) https://doi.org/10.1016/ 0022-2836(84)90309-7

    work page 1984

  43. [43]

    Journal of Cheminformatics 4(1), 17 (2012) https://doi.org/10.1186/1758-2946-4-17

    Hanwell, M.D., Curtis, D.E., Lonie, D.C., Vandermeersch, T., Zurek, E., Hutchison, G.R.: Avogadro: An advanced semantic chemical editor, visualization, and analysis platform. Journal of Cheminformatics 4(1), 17 (2012) https://doi.org/10.1186/1758-2946-4-17

    work page doi:10.1186/1758-2946-4-17 2012

  44. [44]

    SoftwareX 1–2, 19–25 (2015) https://doi.org/10.1016/j.softx.2015.06.001

    Abraham, M.J., Murtola, T., Schulz, R., P´ all, S., Smith, J.C., Hess, B., Lindahl, E.: GROMACS: High perfor- mance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX 1–2, 19–25 (2015) https://doi.org/10.1016/j.softx.2015.06.001

    work page doi:10.1016/j.softx.2015.06.001 2015

  45. [45]

    : Martini 3: a general purpose force field for coarse-grained molecular dynamics

    Souza, P.C.T., Alessandri, R., Barnoud, J., Thallmair, S., Faustino, I., Gr¨ unewald, F., Patmanidis, I., Abdizadeh, H., Bruininks, B.M.H., Wassenaar, T.A., Kroon, P.C., Melcr, J., Nieto, V., Corradi, V., Khan, H.M., Doma´ nski, J., Javanainen, M., Martinez-Seara, H., Reuter, N., Best, R.B., Vattulainen, I., Monticelli, L., Peri- ole, X., Tieleman, D.P., ...

    work page doi:10.1038/s41592-021-01098-3 2021

  46. [46]

    Journal of Chemical Theory and Computation 20(1), 224–238 (2024) https://doi.org/10.1021/acs.jctc.3c01015

    Sasselli, I.R., Coluzza, I.: Assessment of the MARTINI 3 Performance for Short Peptide Self-Assembly. Journal of Chemical Theory and Computation 20(1), 224–238 (2024) https://doi.org/10.1021/acs.jctc.3c01015

    work page doi:10.1021/acs.jctc.3c01015 2024

  47. [47]

    eLife 12 (2025) https://doi

    Kroon, P.C., Grunewald, F., Barnoud, J., Tilburg, M., Brasnett, C., Souza, P.C.T., Wassenaar, T.A., Marrink, S.-J.J.: Martinize2 and Vermouth: Unified Framework for Topology Generation. eLife 12 (2025) https://doi. org/10.7554/eLife.90627.3 17

    work page doi:10.7554/elife.90627.3 2025