bio.rodeo
ModelsOrganizationsProvidersLeaderboardAboutSign in
bio.rodeo

The authoritative source for evaluating biological foundation models. No hype, just honest analysis.

Categories
  • DNA & Gene
  • RNA
  • Protein
  • Small molecule
  • Single-cell
  • Spatial omics
  • Pathology
  • Imaging
  • Metabolomics
  • Biosignals
  • Language model
bio.rodeoModelsOrganizationsProvidersLeaderboardAboutFAQSubmit a modelContact
© 2026 Pulsatance. All rights reserved. ~
Built by Pulsatance
models / protein / hermes
Protein
University of WashingtonInstitute for Protein DesignFred Hutchinson Cancer CenterYale UniversityMITReleased July 2024

HERMES

Structure-based mutational effect prediction from local atomic environments, scoring how substitutions change protein stability and binding affinity.

The short version

  • —Ranks every possible substitution at every site for stability or binding effect
  • —Scores a residue from all atoms within 10 A, including partial charge and solvent exposure
  • —A Zernike Fourier projection makes each score independent of how the structure is oriented
  • —An amortized variant absorbs side-chain repacking, avoiding an explicit relaxation step
  • —Fine-tunes end to end on any measured phenotype instead of fitting a separate regression head
99Openness4Citations
12GitHub stars
MITLicense

Where to run it

No providers recorded yet. Browse all providers

HERMES is a family of structure-based models that predict how an amino acid substitution changes a protein's behaviour, working entirely from the atoms packed around the mutated residue. Rather than reading a sequence or an alignment, it takes a 10 Å all-atom neighbourhood, masks the focal residue, and asks which of the 20 amino acids belongs there. The log-likelihood ratio between the mutant and wild-type amino acid becomes a mutational-effect score, which the authors show tracks changes in thermodynamic folding stability and protein-protein binding affinity without any supervision on those measurements.

The model was developed by Gian Marco Visani, Armita Nourmohammad, and colleagues at the University of Washington Paul G. Allen School of Computer Science and Engineering, with collaborators at the Institute for Protein Design, Fred Hutchinson Cancer Center, Yale University, and MIT. It extends the group's earlier H-CNN architecture with a roughly 2.75x inference speedup and, more consequentially, an architecture that can be fine-tuned end to end for arbitrary phenotypes rather than through a bolted-on regression head.

HERMES occupies the structure-based corner of the mutational-effect landscape, alongside inverse-folding models such as ProteinMPNN and the stability-specialised RaSP, Stability-Oracle, and ThermoMPNN. Its distinguishing move is treating side-chain packing explicitly: three inference protocols differ only in how much the local environment is allowed to rearrange around the substitution.

#Key Features

  • Three inference protocols: HERMES-fixed scores both amino acids on the unmodified wild-type structure; HERMES-relaxed rebuilds the mutant neighbourhood with Rosetta side-chain repacking; HERMES-amortized distills the relaxed protocol into the weights, running at fixed-structure speed.
  • All-atom rotational equivariance: each atom contributes coordinates, element type, partial charge, and solvent-accessible surface area, projected onto a Zernike Fourier basis and processed by SO(3)-equivariant convolutions so scores do not depend on structure orientation.
  • End-to-end phenotype fine-tuning: the model's own predicted score is regressed onto experimental measurements with a Huber loss, calibrating the network itself rather than a downstream head.
  • Antigen stabilisation screening: a dedicated script ranks substitutions that stabilise the metastable pre-fusion conformation of viral envelope proteins, the central objective in structure-based vaccine antigen design.

#Technical Details

HERMES pretrains on masked amino-acid recovery over ProteinNet's CASP12 set, filtered at 30% sequence identity. Each released model is an ensemble of 10 independently trained networks of 3.5M parameters each, with variants trained both on clean coordinates and with 0.5 Å Gaussian coordinate noise; the noised variants score better zero-shot. Because Rosetta side-chain relaxation runs roughly 66x slower than a fixed forward pass, HERMES-amortized is distilled from relaxed predictions on about 15,000 neighbourhoods, 0.5% of the pretraining set.

On zero-shot stabilising-versus-destabilising classification, packing awareness matters: HERMES-relaxed reaches recall 0.48 against 0.27 for HERMES-fixed, with the largest gains on substitutions between residues of very different size. Fine-tuned on the Stability-Oracle cDNA117k and ThermoMPNN Megascale splits, HERMES outperforms RaSP and matches Stability-Oracle and ThermoMPNN, and remains accurate when given ESMFold structures instead of crystallographic ones. On SKEMPI v2.0 binding effects it reaches Spearman 0.286 zero-shot, comparable to ProteinMPNN but behind Rosetta and FoldX at roughly 0.35; fine-tuned on SKEMPI it reaches Pearson 0.471 under a random split and 0.435 under a homology-aware split that separates interaction classes.

#Applications

HERMES suits any workflow that needs to triage a large substitution space quickly: saturation-mutagenesis library design, stability engineering of enzymes and biologics, and interpretation of missense variants where a structure is available. The authors demonstrate a vaccine-design pipeline that screens all substitutions with HERMES-amortized and filters the top hits with Rosetta; in the pre-fusion core workflow, 7 of the top 10 suggested mutations were Rosetta-validated, against 4 for ProteinMPNN and 2 at random. A companion study applies the pretrained model to T-cell receptor recognition, predicting TCR-pMHC binding and T-cell activity at up to 0.72 correlation with experiment and designing immunogenic peptides that activated T-cells at success rates up to 50%.

#Impact

The model's practical appeal is throughput: exhaustive saturation mutagenesis of a whole antigen is cheap enough to run on a single GPU, a scale at which Rosetta screens become impractical. The MIT-licensed repository carries several of the trained ensembles directly in trained_models/ — each seed a multi-megabyte PyTorch checkpoint shipped with its hyperparameters, training log, and loss traces — with the remaining models on Zenodo. A second MIT repository, hermes_experiments, holds the paper's evaluation code, the T2837, Ssym, Megascale, SKEMPI, RaSP, and antigen benchmark tables, and the released prediction matrices, so the reported comparisons can be re-derived rather than taken on trust. The fine-tuning path is documented, so the same architecture can be retargeted to any phenotype for which mutational data exist. Several limitations are worth stating plainly. Scores are computed one site at a time, so synergistic multi-mutation designs are outside the model's reach, and nearby mutated residues bias each other's predictions. Zero-shot binding prediction trails physics-based energy functions. The pretrained models also carry a bias toward size-conserving substitutions, which the amortized and fine-tuned variants substantially reduce. The work remains a preprint and has not been peer reviewed.

At a glance

Parameters
3.5 Million
Released
July 2024
Category
Protein
License
MIT
Organizations
University of Washington / Institute for Protein Design / Fred Hutchinson Cancer Center / Yale University / MIT

Related models

  • HERCULES

    Italian Institute of Technology

  • ProStab

    Shanghai Jiao Tong University

  • METL

    University of Wisconsin-Madison / Morgridge Institute for Research / Duke University

  • DPStab

    National University of Singapore / Central South University

  • SoluProtMut

    Loschmidt Laboratories / International Clinical Research Center (ICRC) / TU Wien / University of Greifswald

Links

GitHub RepositoryGitHub RepositoryGitHub RepositorybioRxiv PreprintDataset

Tags

cnnprotein_designself_supervisedvariant_effect_predictionzero_shot

Something wrong?

Much of this page is generated or calculated automatically. Flag anything that looks off and we will re-run it.