Automated Adaptive Absolute Binding Free Energy Calculations.

Area of Science:

• Computational chemistry
• Drug discovery
• Molecular modeling

Background:

• Alchemical absolute binding free energy (ABFE) calculations are powerful for drug discovery but computationally intensive.
• Existing methods require significant human intervention and computational resources, limiting their widespread application.

Purpose of the Study:

• To develop a fully automated workflow for ABFE calculations to reduce computational cost and human effort.
• To improve the efficiency and robustness of ABFE calculations for practical drug discovery applications.

Main Methods:

• Automated selection of intermediate states (λ windows) with consistent overlap.
• Ensemble-based detection of simulation equilibration using paired t-tests.
• Adaptive allocation of sampling time based on inter-replicate statistics.

Main Results:

• Automated λ window selection is rapid, robust, and simple.
• Paired t-tests provide reliable equilibration detection across multiple runs.
• The automated workflow yields results equivalent to non-adaptive methods, often with accelerated equilibration.

Conclusions:

• The developed automated ABFE workflow significantly enhances efficiency and reduces computational burden.
• This approach makes ABFE calculations more accessible and practical for drug discovery.
• The open-source package A3FE implements this complete workflow for broader use.