Supplementary data for Biomolecular Emulator Overview This repository contains supplementary data for Biomolecular Emulator: Lewis, S.; Hempel, T.; Jiménez-Luna, J.; Gastegger, M.; Xie, Y.; Foong, A. Y. K.; Satorras, V. G.; Abdin, O.; Veeling, B. S.; Zaporozhets, I.; Chen, Y.; Yang, S.; Foster, A. E.; Schneuing, A.; Nigam, J.; Barbero, F.; Stimper, V.; Campbell, A.; Yim, J.; Lienen, M.; Shi, Y.; Zheng, S.; Schulz, H.; Munir, U.; Sordillo, R.; Tomioka, R.; Clementi, C.; Noé, F. Scalable Emulation of Protein Equilibrium Ensembles with Generative Deep Learning. Science 2025. https://doi.org/10.1126/science.adv9817. Supplementary code This repository contains a snapshot of the bioemu inference and benchmark code, bioemu_release.zip: code snapshot of https://github.com/microsoft/bioemu at paper publication date bioemu-benchmarks_release.zip: code snapshot of https://github.com/microsoft/bioemu-benchmarks/ at paper publication date We strongly recommend using the linked GitHub repositories for up-to-date code, installation instructions, and further information. This Zenodo repository represents a static snapshot of supplementary data for the publication. The code is licensed under the MIT License. Model checkpoints The model checkpoints that were used to produce the results for the publication are given in bioemu_finetuned_step_4421376000_epoch_00100.ckpt: Default BioEmu model used throughout the publication desres-models.zip: fast folder model, Fig. 3 A (for developers, please do not use for inference unless explicitly intended) The model checkpoints are licensed under the MIT License. Supplementary structures This repository contains the molecular structures that were used to generate the results in the above publication. The supplementary structures are licensed under Community Data License Agreement - Permissive - Version 2.0. Figure panel data assignment: The samples_*.zip files map to the paper figures as follows: samples_cath.zip: Fig. 3 B, Fig. S10 A samples_cath-scaling-001.zip: Fig. 3 B; Fig. S10 C samples_cath-scaling-010.zip: Fig. 3 B; Fig. S10 C samples_cath-scaling-100.zip: Fig. 3 B; Fig. S10 C samples_delta_G.zip: Fig. 4 B, C and F samples_desres_model.zip: Fig. 3 A; Fig. S9 samples_e2e_multiconf_crypticpocket.zip: Fig. S6 samples_e2e_multiconf_domainmotion.zip: Fig. S6 samples_e2e_multiconf_ood60_local.zip: Fig. S6 samples_e2e_singleconf_localunfolding.zip: Fig. S6 samples_examples_fig3.zip: Fig. 3 C and D; Fig 12 samples_multiconf_crypticpocket.zip: Fig. 2 C; Fig. S4 samples_multiconf_domainmotion.zip: Fig. 2 A; Fig. S2 samples_multiconf_ood60_local.zip: Fig. S1 samples_singleconf_localunfolding.zip: Fig. 2 B; Fig. S3 samples_stable_proteins.zip: Fig 4 D samples_unstable_proteins.zip: Fig 4 E Data loading The data can be loaded with common molecular dynamics tools, e.g.,with mdtraj: import mdtrajsamples = mdtraj.load("path/to/samples.xtc", top="path/to/topology.pdb")