Understanding the thermodynamics and spectroscopic properties of biomolecules hinges on accurately modelling their energy landscapes and vibrational behaviour. Sergio Suárez‑Dou, Miguel Gallegos, and Kyunghoon Han, alongside colleagues at the University of Luxembourg, demonstrate a new force field, SO3LR, capable of reproducing high‑level density‑functional theory calculations with remarkable accuracy. The researchers validated SO3LR across a diverse range of bio‑relevant molecules, extending far beyond its original training data and successfully predicting vibrational characteristics such as frequencies and infrared spectra. Detailed simulations of amino acids, peptides, and protein domains—both in vacuum and aqueous environments—reveal that SO3LR consistently matches the potential‑energy surfaces generated by density‑functional theory, capturing crucial effects like anharmonicity and environmental interactions. This work establishes that machine‑learning force fields can deliver accuracy comparable to quantum‑mechanical methods, but at a significantly reduced computational cost, thereby enabling more extensive and detailed studies of biomolecular systems.

Predicting biomolecular thermodynamics and spectroscopy requires accurate determination of the relative energies of metastable states and the curvature of the potential‑energy surface. Researchers have demonstrated that the general‑purpose SO3LR machine‑learned force field reproduces density‑functional theory calculations (specifically the PBE0+MBD method) with unprecedented accuracy across bio‑relevant molecules. This performance extends to systems larger and more complex than those used to train the force field, indicating a strong ability to generalise to new scenarios. For a set of 23 small molecules, SO3LR accurately captures harmonic and anharmonic vibrational characteristics, including frequencies, displacement patterns, and infrared spectra.

Detailed dynamical studies were then performed on the o‑Phe⁺ amino acid, the alanine‑15 peptide, and the assembly of p53 transactivation domains into tetramers, both in vacuum and aqueous environments. Measurements confirm that SO3LR consistently matches results obtained using density‑functional theory, providing a quantum‑accurate picture of metastable minima and vibrational properties at a significantly reduced computational cost. The breakthrough delivers a method for accurately modelling complex biomolecular systems, achieving density‑functional theory accuracy with the efficiency of force‑field calculations. This is achieved through SO3LR’s all‑body treatment of interactions, extending up to approximately 15 Å, enabling the capture of intricate structural details and environmental effects.

SO3LR Force Field Achieves High‑Fidelity Simulations

Scientists have achieved unprecedented fidelity in biomolecular simulations by demonstrating the accuracy of the SO3LR force field against density‑functional theory calculations. The research team meticulously validated SO3LR across a diverse set of bio‑relevant molecules, extending far beyond the scope of its original training data. Experiments revealed that SO3LR accurately reproduces potential‑energy surfaces, vibrational densities of states, and mode eigenvectors, effectively capturing crucial anharmonicity, polarization, and medium‑range interactions essential for understanding protein behaviour. For a cohort of 23 small molecules, SO3LR precisely captured harmonic and anharmonic vibrational features, including frequencies, displacement patterns, and complete infrared spectra.

Detailed dynamical studies were then performed on the o‑Phe⁺ amino acid, the alanine‑15 peptide, and the assembly of p53 transactivation domains into tetramers, both in vacuum and aqueous environments. Measurements confirm that SO3LR consistently matches density‑functional theory results, providing a quantum‑accurate picture of metastable minima and vibrational properties at a significantly reduced computational cost. Data show that SO3LR’s all‑body treatment of interactions, extending up to approximately 15 Å, enables the capture of intricate structural details and environmental effects.

This capability is crucial for modelling polarization, charge transfer, and exchange repulsion—phenomena often neglected in conventional force fields. Results demonstrate that machine‑learned force‑field‑driven dynamics, such as those powered by SO3LR, provide a robust framework for investigating biomolecular stability and vibrations from first principles. The study establishes that SO3LR’s ability to account for medium‑range interactions and delicate electronic effects significantly improves the accuracy of simulations. This advancement opens new possibilities for the computational study of biomolecules, promising deeper insights into their structure, dynamics, and function.

SO3LR Matches DFT Accuracy for Biomolecules

The research demonstrates the exceptional accuracy of the SO3LR force field in reproducing density‑functional theory calculations (using the PBE0+MBD method) across a diverse range of biomolecules. The force field accurately predicts harmonic and anharmonic vibrational features—including frequencies and infrared spectra—for small molecules, and successfully models the dynamics of more complex systems like peptides and protein domains both in vacuum and in aqueous solution. This achievement represents a significant step toward accurately simulating biomolecular behaviour using computationally efficient methods. The study establishes that dynamics driven by SO3LR provide a level of accuracy comparable to density‑functional theory calculations, but at a substantially reduced computational cost.

The advancement opens opportunities for detailed investigations of biomolecular stability and vibrations, extending to systems previously inaccessible due to computational limitations, such as intrinsically disordered proteins and those containing non‑natural amino acids. The authors acknowledge limitations inherent in the PBE0+MBD reference method, particularly potential overestimation of charge‑transfer contributions in certain systems. Future work will likely focus on extending the application of SO3LR to even larger and more complex biomolecular assemblies, further refining its accuracy and exploring its potential for predictive molecular‑dynamics simulations.

---

More information

Stability and Vibrations of Proteins in Vacuum and Water: Bridging Quantum Accuracy and Force‑Field Efficiency

ArXiv: https://arxiv.org/abs/2601.09845