Toward empirical force fields that match experimental observables.

Area of Science:

• Computational Chemistry
• Biophysics
• Molecular Modeling

Background:

• Biomolecular force fields traditionally rely on limited quantum chemistry data and small fragment experiments.
• Advances in computational power enable extensive molecular dynamics simulations of large systems with ergodic sampling.
• Solution experiments on macromolecular systems provide valuable data for parameterization.

Purpose of the Study:

• To review recent automated methods for deriving biomolecular force fields directly from simulations and macromolecular experiments.
• To highlight the connection between these novel methods and machine learning approaches.
• To discuss current challenges and future directions in the field.

Main Methods:

• Review of automated methodologies for force field derivation.
• Analysis of integration strategies for simulation data and solution scattering experiments.
• Exploration of machine learning techniques applied to force field parameterization.

Main Results:

• Emergence of automated methods capable of leveraging extensive simulation and experimental data.
• Demonstration of the synergy between simulation-based approaches and machine learning.
• Identification of key challenges in achieving robust and accurate force fields through these new paradigms.

Conclusions:

• Direct derivation of biomolecular force fields from large-scale simulations and macromolecular experiments is becoming feasible.
• Automated methods and machine learning are crucial for advancing this approach.
• Further research is needed to address open challenges in the field for broader applicability.