EUVIMEDOpdag EUVIMED
Open access · CC BY-NC-ND 4.0

DreamFold : A World Model to efficiently generate protein folding pathways in the latent space

Alan Ianeselli, Jewon Im iD, Eddie Cavallin iD, Mark B Gerstein iD

Protein Science · 2026-11 · 35 (11)

DOI: 10.1002/pro.70794 ↗

At a glance

Journal article in Protein Science, published online on 2026-09-30. Written by Alan Ianeselli and 3 co-authors. Institution: Department of Molecular Biophysics and Biochemistry, Yale University, New Haven, Connecticut, USA. Major MEDLINE topics: Protein Folding, Proteins, Neural Networks, Computer. 73 references. The full text is freely available (open access, CC BY-NC-ND 4.0). Funded by: NIH HHS, Yale University, NIH.

AbstractSubjectsBibliografiske oplysningerCiteRelated studies

Abstract (original language)

While there is an abundance of static data for the structure of biological macromolecules, the data regarding their folding mechanisms and dynamics is scarce, posing a challenge to the training of AI models. The World Model approach, which has been very successful in robotics, can come in handy in this regard. From limited data, it can build a latent, approximated representation of the spatio‐temporal folding environment that can be used to train downstream AI models. Here, we developed DreamFold, a World Model‐based generative framework to perform biomolecular simulations. DreamFold encodes protein structures and dihedral angle moves into latent vectors using variational autoencoders. Then, it uses another neural network to rapidly predict the next latent structure after a latent move, allowing the model to develop its own understanding of protein dynamics. Finally, in a “hallucinated” latent environment, an agent learns, through evolutionary algorithms, a policy to drive folding simulations toward a target. Latent configurations can then be decoded back into atomistic structures. DreamFold can compute protein folding pathways four orders of magnitude faster than molecular dynamics (MD), up to ~30,000×. Furthermore, for the monomeric proteins tested here (up to 449 aa), the computational cost scales linearly with the number of atoms ( N ) as O( N ) (in comparison, MD scales as O( N log N )). We have validated our results against established MD benchmarks and other available experimental data. Finally, we show how DreamFold allows us to identify folding intermediates with cryptic binding sites that are therapeutically valuable. Overall, DreamFold facilitates the study of protein dynamics by generative AI with applications in structure‐based drug discovery.

Subjects

MeSH terms (MEDLINE)

Protein FoldingProteinsNeural Networks, ComputerModels, MolecularAutoencoderProtein ConformationMolecular Dynamics SimulationAlgorithms

Author keywords

Artificial Intelligencedrug discoverylatent spaceprotein foldingworld models

Substances

Proteins

Authors and institutions

  • Department of Molecular Biophysics and Biochemistry, Yale University, New Haven, Connecticut, USA.
  • Department of Computer Science, Yale University, New Haven, Connecticut, USA.
  • Department of Statistics and Data Science, Yale University, New Haven, Connecticut, USA.
  • Program in Computational Biology and Bioinformatics, Yale University, New Haven, Connecticut, USA.
  • Department of Biomedical Informatics & Data Science, Yale University, New Haven, Connecticut, USA.

ORCID profiles: Jewon Im, Eddie Cavallin, Mark B Gerstein

Funding

  • NIH HHS
  • Yale University
  • NIH

Cite

APA

Ianeselli, A., Im, J., Cavallin, E., & Gerstein, M. B. (2026). DreamFold : A World Model to efficiently generate protein folding pathways in the latent space. Protein Science, 35(11). https://doi.org/10.1002/pro.70794

Vancouver

Ianeselli A, Im J, Cavallin E, Gerstein MB. DreamFold : A World Model to efficiently generate protein folding pathways in the latent space. Protein Science. 2026;35(11). doi:10.1002/pro.70794

BibTeX
@article{Ianeselli2026,
  author = {Ianeselli, Alan and Im, Jewon and Cavallin, Eddie and Gerstein, Mark B},
  title = {DreamFold : A World Model to efficiently generate protein folding pathways in the latent space},
  journal = {Protein Science},
  year = {2026},
  volume = {35},
  number = {11},
  doi = {10.1002/pro.70794}
}
Start for free →