aiqbench analysis
This research introduces EnsembleEGNN, a foundation model that encodes conformational ensembles of cyclic peptides using shared Equivariant Graph Neural Networks to improve property prediction accuracy. By pretraining on multi-task self-supervised objectives and integrating sequence information via co-training with BERT, the method significantly outperforms single-conformer baselines in modeling molecular thermodynamics. The work advances machine learning applications for biomolecules by demonstrating that explicit ensemble representation yields more robust embeddings than traditional representative structures.
Graph Learning on Ensembles of Cyclic Peptides: An Investigation of Molecular Ensemble Modeling
arXiv:2607.21561v1 Announce Type: new Abstract: Molecular property prediction from structure often uses a single representative conformation, even though many molecules exist as conformational ensembles in solution. We introduce EnsembleEGNN, a molecular ensemble foundation model that encodes an ensemble by first encoding each conformer with shared Equivariant Graph Neural Network (EGNN) layers, then pooling the resulting conformer representations with a Set Attention Block. We pretrain the model on CREMP, a cyclic peptide ensemble dataset, using a multi-task self-supervised objective combining masked token recovery, noisy-coordinate reconstruction, and pairwise distance reconstruction. On the CREMP-CycPeptMPDB dataset, training EnsembleEGNN from scratch fails entirely ($R^2=0.005$). However, the pretrained model reaches $R^2=0.477$ and Pearson $r=0.699$, outperforming the sequence-only BERT baseline ($R^2=0.439$, Pearson $r=0.667$). When EnsembleEGNN is co-trained end-to-end with the BERT sequence encoder, the hybrid model improves further to $R^2=0.538$ and Pearson $r=0.737$. These results demonstrate that encoding conformational ensembles into a single thermodynamically informed embedding improves cyclic-peptide property prediction.
Abstract:Molecular property prediction from structure often uses a single representative conformation, even though many molecules exist as conformational ensembles in solution. We introduce EnsembleEGNN, a molecular ensemble foundation model that encodes an ensemble by first encoding each conformer with shared Equivariant Graph Neural Network (EGNN) layers, then pooling the resulting conformer representations with a Set Attention Block. We pretrain the model on CREMP, a cyclic peptide ensemble dataset, using a multi-task self-supervised objective combining masked token recovery, noisy-coordinate reconstruction, and pairwise distance reconstruction. On the CREMP-CycPeptMPDB dataset, training EnsembleEGNN from scratch fails entirely ($R^2=0.005$). However, the pretrained model reaches $R^2=0.477$ and Pearson $r=0.699$, outperforming the sequence-only BERT baseline ($R^2=0.439$, Pearson $r=0.667$). When EnsembleEGNN is co-trained end-to-end with the BERT sequence encoder, the hybrid model improves further to $R^2=0.538$ and Pearson $r=0.737$. These results demonstrate that encoding conformational ensembles into a single thermodynamically informed embedding improves cyclic-peptide property prediction.
Accepted to Graph Foundation Models workshop at ICML '26. Contains 8 pages, 4 figures
Machine Learning (cs.LG); Biomolecules (q-bio.BM)
arXiv:2607.21561 [cs.LG]
(or arXiv:2607.21561v1 [cs.LG] for this version)
https://doi.org/10.48550/arXiv.2607.21561 Focus to learn more arXiv-issued DOI via DataCite (pending registration)
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Full text under the publisher's open licence · 258 words
Related coverage
Transferable FB-GNN-MBE Framework for Potential Energy Surfaces: Data-Adaptive Transfer Learning in Deep Learned Many-Body Expansion Theory
arXiv · Jul 28, 2026
Improving synthesis prediction of small molecules at scale with RetroChimera
Microsoft Research · Sep 21, 2026
Boltzmann-Expected Molecular Design with Decoupled Annealing Flows
arXiv · Jul 23, 2026
Semi-Supervised Learning for Molecular Graphs via Ensemble Consensus
arXiv · Jul 31, 2026