Extended Adaptive Biasing Force Algorithm. An On-the-Fly Implementation for Accurate Free-Energy Calculations.

Area of Science:

• Computational Chemistry
• Molecular Dynamics
• Biophysics

Background:

• Traditional adaptive biasing force (ABF) algorithms require specific prerequisites, limiting their application in free-energy calculations.
• Extended ABF (eABF) addresses these limitations but typically necessitates post-hoc analysis, reducing reliability and efficiency.
• Existing eABF implementations in NAMD require trajectory post-processing, leading to information loss and computational overhead.

Purpose of the Study:

• To develop and validate a user-friendly, on-the-fly extended adaptive biasing force (eABF) code for NAMD.
• To compare the performance of the on-the-fly eABF algorithm against traditional ABF and post-hoc eABF.
• To provide guidelines for parameter selection in the on-the-fly eABF implementation.

Main Methods:

• Development of an integrated, on-the-fly eABF simulation code within the NAMD molecular dynamics engine.
• Testing the new code across eight diverse illustrative examples.
• Comparative analysis of free-energy landscapes and computational effort against traditional ABF and post-hoc eABF.

Main Results:

• The on-the-fly eABF algorithm accurately determines free-energy landscapes for systems with coupled variables or restraints.
• It significantly enhances the reliability of free-energy estimates compared to post-hoc analysis.
• The computational cost of the on-the-fly eABF is comparable to regular ABF, with negligible overhead.

Conclusions:

• The developed on-the-fly eABF algorithm represents a significant advancement over previous ABF methods.
• It offers improved accuracy, reliability, and efficiency for free-energy calculations in complex molecular systems.
• This 'second generation' ABF algorithm is poised for widespread use in investigating recognition and association phenomena.