EnsembleEGNN: Set-Based Graph Learning for Thermodynamic Ensembles of Cyclic Peptides
Abstract
Molecular graph encoding often relies on a single, static structure, ignoring the thermodynamic ensemble of molecules that are present in solution. Here, we introduce EnsembleEGNN, a foundation model that encodes structural ensembles by processing individual conformers through shared equivariant graph neural network layers, pooled with a set attention block, to make property predictions from the whole ensemble. Pretrained on the CREMP cyclic peptide dataset using multi-task self-supervision, the model is trained to encode the conformational variability of each molecule. When predicting membrane permeability from the CycPeptMPDB benchmark, EnsembleEGNN achieves an R2 of $0.477$ under random cross-validation, outperforming a sequence-only BERT baseline (R2=0.439). This representation advantage persists under rigorous out-of-distribution Butina splits (R2=0.401 versus $0.354$). Finally, a hybrid architecture co-training EnsembleEGNN with the BERT model achieves the highest overall accuracy across both random (R2=0.538) and structural holdout evaluations (R2=0.444). These results demonstrate that encoding conformational ensembles into latent representations improves predictions for properties governed by thermodynamics.
Related
- Generation of membrane-permeable cyclic peptides inhibiting protein–protein interaction
- Comprehensive Strategies in the Synthesis of Noncanonical Cyclic Peptides Containing Biaryl-Bridged Linkages
- DeepEnzyKat: Turnover number prediction through enzyme-substrate interaction modeling and adaptive expert ensembles
- Functionalization of Peptides via Radical Arylation
- Cardiolipin regulates free energy of folding and vesicle leakage of CM15 antimicrobial peptides in bacterial inner membranes
BibTeX
@misc{feller2026ensembleegnn,
title = {{EnsembleEGNN: Set-Based Graph Learning for Thermodynamic Ensembles of Cyclic Peptides}},
author = {Aaron L. Feller and Kris Deibler and Maxim Secor},
howpublished = {arXiv},
year = {2026},
month = sep,
doi = {10.48550/arxiv.2607.21561},
eprint = {2607.21561},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2607.21561},
note = {CC BY 4.0},
}
RIS
TY - UNPB
TI - EnsembleEGNN: Set-Based Graph Learning for Thermodynamic Ensembles of Cyclic Peptides
AU - Aaron L. Feller
AU - Kris Deibler
AU - Maxim Secor
PY - 2026
DA - 2026/09/30
PB - arXiv
DO - 10.48550/arxiv.2607.21561
UR - https://arxiv.org/abs/2607.21561
AB - Molecular graph encoding often relies on a single, static structure, ignoring the thermodynamic ensemble of molecules that are present in solution. Here, we introduce EnsembleEGNN, a foundation model that encodes structural ensembles by processing individual conformers through shared equivariant graph neural network layers, pooled with a set attention block, to make property predictions from the whole ensemble. Pretrained on the CREMP cyclic peptide dataset using multi-task self-supervision, the model is trained to encode the conformational variability of each molecule. When predicting membrane permeability from the CycPeptMPDB benchmark, EnsembleEGNN achieves an R^2 of $0.477$ under random cross-validation, outperforming a sequence-only BERT baseline (R^2=0.439). This representation advantage persists under rigorous out-of-distribution Butina splits (R^2=0.401 versus $0.354$). Finally, a hybrid architecture co-training EnsembleEGNN with the BERT model achieves the highest overall accuracy across both random (R^2=0.538) and structural holdout evaluations (R^2=0.444). These results demonstrate that encoding conformational ensembles into latent representations improves predictions for properties governed by thermodynamics.
ER -