CONFPASS: Fast DFT Re-Optimizations of Structures from Conformation Searches

Area of Science:

• Computational Chemistry
• Molecular Modeling
• Drug Discovery

Background:

• Accurate conformational analysis is crucial for predicting molecular properties.
• Density functional theory (DFT) re-optimization of conformers is computationally intensive.
• Current methods for prioritizing conformers can lead to redundant calculations and may miss global minima.

Purpose of the Study:

• To develop and evaluate CONFPASS (Conformer Prioritizations and Analysis for DFT re-optimizations).
• To provide a confidence estimate for finding the global minimum structure.
• To reduce duplication rates in DFT re-optimization workflows.

Main Methods:

• CONFPASS extracts dihedral angle descriptors from conformational search outputs.
• It performs clustering and generates a priority list for DFT re-optimizations.
• Evaluations used DFT data for 150 diverse, flexible molecules.

Main Results:

• CONFPASS provides a confidence estimate for global minimum identification.
• 90% confidence in finding the global minimum is achieved after optimizing half of the force field (FF) structures.
• CONFPASS reduces duplication rates by a factor of 2 for the first 30% of re-optimizations, capturing the global minimum 80% of the time.

Conclusions:

• CONFPASS is an effective tool for prioritizing molecular conformers for DFT re-optimization.
• The software enhances computational efficiency by minimizing redundant calculations.
• CONFPASS increases the likelihood of identifying the global minimum structure early in the re-optimization process.