The physical foundation beneath protein generative modeling
Abstract
Physical equations in protein modeling appear to have been replaced by generative models trained directly on structure data. By learning a mapping from noise to data, sampling de novo protein structures has become much more efficient. However, such models can also learn non-physical features and break down with out-of-distribution settings which are typical in protein design campaigns. In this review, we highlight where the physics persists in protein generative modeling pipelines and the issues that linger when physically relevant components of a macromolecular system are left unmodeled. We present a perspective that respecting the underlying physics of macromolecular systems, increasingly through learned representations that are physically grounded and updating generative models with experimental data, is foundational to generative modeling for functional protein design.
Related
- Generative AI for controllable protein sequence design: A survey
- DNA-BFN: Bayesian flow networks for generative modeling of regulatory DNA sequences
- Aligning protein-generative models to experimental fitness with ProteinDPO
- A Generative Neuro‐Symbolic AI for Protein Sequence Design
- CardamomOT: A mechanistic optimal transport-based framework for gene regulatory network inference, trajectory reconstruction and generative modeling
BibTeX
@misc{lu2026physical,
title = {{The physical foundation beneath protein generative modeling}},
author = {Tianyu Lu and Po-Ssu Huang},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2609.04465},
eprint = {2609.04465},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.04465},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - The physical foundation beneath protein generative modeling
AU - Tianyu Lu
AU - Po-Ssu Huang
PY - 2026
DA - 2026/09/03
PB - arXiv
DO - 10.48550/arxiv.2609.04465
UR - https://arxiv.org/abs/2609.04465
AB - Physical equations in protein modeling appear to have been replaced by generative models trained directly on structure data. By learning a mapping from noise to data, sampling de novo protein structures has become much more efficient. However, such models can also learn non-physical features and break down with out-of-distribution settings which are typical in protein design campaigns. In this review, we highlight where the physics persists in protein generative modeling pipelines and the issues that linger when physically relevant components of a macromolecular system are left unmodeled. We present a perspective that respecting the underlying physics of macromolecular systems, increasingly through learned representations that are physically grounded and updating generative models with experimental data, is foundational to generative modeling for functional protein design.
ER -