Acellera AceFF-2-OMOL

Organization(s): Acellera Therapeutics, Inc.
Contact: info@acellera.com
License: Apache 2.0

This model was presented in the paper AceFF: A State-of-the-Art Machine Learning Potential for Small Molecules.

---

Overview

Acellera AceFF-2-OMOL is a next-generation machine learning interatomic potential (MLIP) designed for small molecules. It addresses key limitations of traditional molecular mechanics (MM) force fields and earlier NNP models, including restricted atom types, limited charge support, and computational inefficiencies.

The model leverages the TensorNet v2 architecture [Farr2026], an evolution of TensorNet [Simeon2024]. It is implemented in the MLIP software library TorchMD-Net [Pelaez2024, Farr2026] to provide accurate predictions for diverse drug-like compounds, supporting all key chemical elements and charged molecules. Acellera AceFF-2-OMOL is part of the AceFF family of potentials released by Acellera. Unlike AceFF-2.0, which was trained on Acellera's internal proprietary dataset, AceFF-2-OMOL is trained on the public OMol25 dataset [Levine2025] of molecular forces and energies computed at the wB97M-V/def2-tzvppd level of theory. We kept only structures containing the AceFF-2.0 elements plus sodium and potassium: H, B, C, N, O, F, Na, Si, P, S, Cl, K, Br, and I.

---

Benchmarks

Wiggle150

The table shows the results on the Wiggle150 benchmark. We include AIMNet2 and ANI-2x for comparison.

Method	MAE (kcal/mol)	RMSE (kcal/mol)
AceFF-2-OMOL	1.25	1.64
AceFF-2.0	1.76	2.34
AceFF-1.1	2.51	3.18
AceFF-1.0	2.73	3.32
AIMNet2	2.39	3.13
ANI-2X	4.41	5.41

Performance of MLIPs on Wiggle150 benchmark

Many more benchmarks are reported in the paper, including torsion scans and speed.

---

Key Features

• Broad Applicability: Supports diverse drug-like molecules, including charged species and rare chemical groups.
• High Accuracy: Benchmarked on a wide array of test cases.
• Improved Stability: Enables a 2 fs timestep for NNP/MM simulations, significantly reducing computational costs.
• Integration-Friendly: Available for RBFE calculations via HTMD.
• Open Science: The model is available with an Apache 2.0 license.

---

Usage

1. Example notebooks are available in Google Colab, demonstrating the use of Acellera AceFF with OpenMM and ASE.

• Single point energy calculator with ASE.
• Batch minimization and molecular dynamics with PyTorch.
• Running MD for a small molecule in vacuum with OpenMM.
• Running MD for a protein-ligand complex using MM/ML mechanical embedding scheme with OpenMM.

2. Run ML potential molecular simulations of a small molecule using ACEMD with this tutorial, e.g., to minimize.
3. For a tutorial on running mixed protein-ligand simulations, refer to NNP/MM in ACEMD.

---

Applications

• Drug Discovery: Optimizing lead compounds in hit-to-lead and lead optimization stages using free energy methods.
• Binding Free Energy Calculations: Accurate and efficient RBFE predictions for diverse molecular systems [Zariquiey2025].
• Molecular dynamics: Capturing higher-body terms than traditional MM force fields, Acellera AceFF can be used for structure minimization and dynamics of small molecules.

Limitations

• Supported elements only: Acellera AceFF-2-OMOL is trained on the subset of OMol25 containing only the elements H, B, C, N, O, F, Na, Si, P, S, Cl, K, Br, and I.
• Time step: Use time steps of 2 fs to run dynamics with hydrogen mass repartitioning.

---

References

[Simeon2024] Simeon, Guillem, and Gianni De Fabritiis, TensorNet: Cartesian tensor representations for efficient learning of molecular potentials, Advances in Neural Information Processing Systems 36 (2024), https://arxiv.org/abs/2306.06482

[Pelaez2024] Raul P. Pelaez, Guillem Simeon, Raimondas Galvelis, Antonio Mirarchi, Peter Eastman, Stefan Doerr, Philipp Thölke, Thomas E. Markland, Gianni De Fabritiis, TorchMD-Net 2.0: Fast Neural Network Potentials for Molecular Simulations, J. Chem. Theory Comput. 2024, 20, 10, 4076–4087, https://arxiv.org/abs/2402.17660

[Zariquiey2025] Francesc Sabanés Zariquiey, Stephen E. Farr, Stefan Doerr, Gianni De Fabritiis, QuantumBind-RBFE: Accurate Relative Binding Free Energy Calculations Using Neural Network Potentials, https://arxiv.org/abs/2501.01811 (2025).

[Levine2025] Daniel S. Levine, Muhammed Shuaibi, Evan Walter Clark Spotte-Smith, et al., The Open Molecules 2025 (OMol25) Dataset, Evaluations, and Models, https://arxiv.org/abs/2505.08762 (2025).

[Farr2026] Stephen E. Farr, Stefan Doerr, Antonio Mirarchi, Francesc Sabanes Zariquiey, Gianni De Fabritiis, AceFF: A State-of-the-Art Machine Learning Potential for Small Molecules, https://arxiv.org/abs/2601.00581