Development and Benchmarking of Open Force Field 2.0.0: The Sage Small Molecule Force Field.

Area of Science:

• Computational chemistry
• Molecular modeling
• Drug discovery

Background:

• The Open Force Field (OpenFF) initiative develops open-source small molecule force fields.
• Previous iterations like Parsley have been foundational for molecular simulations.
• Accurate force fields are crucial for predicting molecular behavior and interactions.

Purpose of the Study:

• Introduce OpenFF 2.0.0 (Sage), a new small molecule force field for drug-like molecules.
• Detail the methodology, innovations, and improvements over previous OpenFF versions.
• Validate Sage's performance in protein-ligand simulations and physical property predictions.

Main Methods:

• Developed Sage based on direct chemical perception and substructure queries.
• Retrained Lennard-Jones (LJ) parameters using condensed phase mixture data.
• Refitted valence parameters using an expanded quantum chemical calculation database and improved fitting procedures.
• Validated against AMBER biopolymer force fields for protein-ligand simulations.

Main Results:

• Sage demonstrates improved performance metrics against quantum chemistry data (RMSD, TFD, ΔΔE).
• Sage shows enhanced accuracy in predicting physical properties (ΔHmix, ρ(x), ΔGsolv, ΔGtrans) compared to experimental data.
• Protein-ligand binding free energy (ΔGbind) predictions with Sage are statistically similar to previous force fields.
• Sage is compatible with AMBER biopolymer force fields.

Conclusions:

• OpenFF 2.0.0 (Sage) represents a significant advancement in small molecule force fields.
• Sage provides improved accuracy for molecular simulations and physical property predictions.
• The open availability of Sage and its training data promotes reproducibility and further research.