Biomolecular Multiscale Simulation (BMS25) Dataset

Training Set for Neural Network Potentials for Electrostatic-Embedding QM/MM Settings


Loading...

Date

2025-12-08

Publication Type

Dataset

ETH Bibliography

yes

Citations

Altmetric

Abstract

Using an accelerated sampling approach, we generated the biomolecular multiscale simulation (BMS25) dataset with nearly 60'000 topologies and more than 1.5 million unique conformations of peptides and miniproteins as well as small molecules and transition states from chemical reactions. The dataset includes energies, gradients, and multipoles of solute molecules as well as gradients on solvent molecules at the wB97M-D4/ma-def2-TZVPP level of theory, enabling the development of multiscale neural network potentials for simulating large biomolecular systems.

Publication status

External links

Editor

Contributors

Contact person: Riniker, Sereina
Data collector: Thürlemann, Moritz
Data collector : Pultar, Felix
Data collector : Gordiy, Igor
Project leader: Riniker, Sereina
Research group: Riniker, Sereina

Book title

Journal / series

Volume

Pages / Article No.

Publisher

ETH Zurich

Event

Edition / version

2025

Methods

Software

ORCA (5.0.4) OpenMM (8.0)

Geographic location

Date collected

Date created

Subject

Computational chemistry; Quantum chemistry; Machine learning

Organisational unit

02020 - Dep. Chemie und Angewandte Biowiss. / Dep. of Chemistry and Applied Biosc.

Notes

Funding

212732 - Combining Molecular Dynamics and Machine Learning for Free Energy Calculation with Quantum-Mechanical Accuracy (SNF)

Related publications and datasets