FeNNix-Bio1

An efficient machine-learning interatomic potential for molecular dynamics simulations of organic and biological systems trained on an extension of the SPICE2 dataset. The model comes in two sizes (S and M).

Model Details

Uses

  • Molecular dynamics simulations of organic and biological systems
  • Geometry optimization
  • Scoring (Binding free energy calculations)

How to Get Started with the Model

Download one of the models from github.com/FeNNol-tools/FeNNol-PMC/tree/main/FENNIX-BIO1. Install the FeNNol library

pip install fennol[cuda]

The model can easily be used with ASE:

from fennol.ase import FENNIXCalculator
import sys
from ase import Atoms
from ase.md.verlet import VelocityVerlet
from ase.io import read
from ase import units
import time

xyz_file = sys.argv[1]
atoms = read(xyz_file)
calc = FENNIXCalculator(model="fennix-bio1M.fnx", gpu_preprocessing=True)
atoms.calc = calc

dyn = VelocityVerlet(atoms, 1.0 * units.fs)  # 1 fs time step
nsteps = 100_000
steps_per_print = 100
t0 = time.time()
for step in range(1,nsteps+1):
    dyn.run(1)
    if step % steps_per_print == 0:
        t1 = time.time()
        steps_per_second = steps_per_print / (t1 - t0)
        print(f"Step: {step:10}, Energy: {atoms.get_potential_energy():15.5f} eV, Steps/s: {steps_per_second:.2f}")
        t0 = time.time()

For better performance, we recommend directly using FeNNol's MD engine (see the example directory in FeNNol's github repo)

Citation

Plé T, Adjoua O, Benali A, Posenitskiy E, Villot C, Lagardère L, Piquemal J-P A Foundation Model for Accurate Atomistic Simulations in Drug Design. ChemRxiv. 2025; doi:10.26434/chemrxiv-2025-f1hgn-v4

BibTeX:

@article{Ple-2025,
  title={A Foundation Model for Accurate Atomistic Simulations in Drug Design},
  DOI={10.26434/chemrxiv-2025-f1hgn-v4},
  journal={ChemRxiv},
  author={Plé, Thomas and Adjoua, Olivier and Benali, Anouar and Posenitskiy, Evgeny and Villot, Corentin and Lagardère, Louis and Piquemal, Jean-Philip},
  year={2025}
}
Downloads last month

-

Downloads are not tracked for this model. How to track
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support