Flash-Molecular-Dynamics
Collection
1 item • Updated
CGSchNet: Coarse-Grained SchNet Model
Pre-trained coarse-grained SchNet neural network force field for molecular dynamics simulation of proteins. Exactly copied from CGSchNet [Charron et. al. 2025].
Model
model_and_prior.pt- CGSchNet model with physical priors (harmonic bonds/angles, dihedrals, repulsion). Trained on 12 proteins with 5-bead-per-residue coarse-graining.
Initial Configurations
Initial configurations for each protein, located in configurations/:
| Protein | File | Description |
|---|---|---|
| 1ENH | 1enh_configurations.pt |
Engrailed homeodomain (54 residues) |
| CLN | cln_configurations.pt |
Chignolin (10 residues) |
| 1YRF | 1yrf_configurations.pt |
WW domain (35 residues) |
| 2A3D | 2a3d_configurations.pt |
Three-helix bundle (73 residues) |
| 2JOF | 2jof_configurations.pt |
Trp-cage (20 residues) |
| 2NUZ | 2nuz_configurations.pt |
Alpha-beta protein (57 residues) |
| 2CI2 | 2ci2_configurations.pt |
CI2 (65 residues) |
| 1RIS | 1ris_configurations.pt |
Ribosomal protein (104 residues) |
| 1FME | 1fme_configurations.pt |
Villin headpiece (114 residues) |
| 3ZBE | 3zbe_configurations.pt |
Ubiquitin-like (304 residues) |
| OPEP-0015 | opep_0015_configurations.pt |
OPEP peptide |
| OPEP-0034 | opep_0034_configurations.pt |
OPEP peptide |
Additional multi-chain configurations: 1d3z_with_ext_PUMA_configurations.pt, mcl1_with_ext_PUMA_configurations.pt, ext_PUMA_alone_configurations.pt, 1enh_elongated_configurations.pt, 2a3d_elongated_configurations.pt.
PDB Structures
Coarse-grained PDB structures (5 beads per residue) in structures/.
Usage with FlashMD
from flashmd.hub import from_pretrained, download_file
# Load model
model = from_pretrained("pingzhili/cg-schnet")
# Download configurations
configs = download_file("pingzhili/cg-schnet", "configurations/1enh_configurations.pt")
References
Transferable coarse-grained model:
Charron, N.E., Bonneau, K., Pasos-Trejo, A.S. et al. Navigating protein landscapes with a machine-learned transferable coarse-grained model. Nat. Chem. 17, 1284–1292 (2025). https://doi.org/10.1038/s41557-025-01874-0.
FlashMD acceleration:
Li, P., Li, H., Liu, Z., Lin, X., & Chen, T. FlashSchNet: Fast and Accurate Coarse-Grained Neural Network Molecular Dynamics. arXiv:2602.13140 (2026). https://arxiv.org/abs/2602.13140
License
MIT
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support