Crystal structure prediction of flexible molecules using parallel genetic algorithms with a standard force field

Area of Science:

• Computational chemistry
• Crystallography
• Materials science

Background:

• Crystal structure prediction (CSP) is crucial for understanding material properties.
• Predicting structures of flexible molecules presents unique challenges.
• Accurate force fields and efficient sampling methods are essential for CSP.

Purpose of the Study:

• To apply the modified genetic algorithms for crystal and cluster prediction (MGAC) framework for flexible molecule CSP.
• To evaluate the performance of MGAC using the general Amber force field (GAFF) and CHARMM.
• To identify areas for improvement in the CSP methodology.

Main Methods:

• Utilized a distributed computing framework integrating multiple programs.
• Employed a parallel genetic algorithm for crystal structure sampling.
• Incorporated local energy minimization and structure classification.

Main Results:

• The MGAC framework successfully predicted experimentally known crystal structures for flexible molecules.
• Some instances showed missing structures and suboptimal ranking, indicating limitations.
• The general Amber force field (GAFF) and CHARMM were used in conjunction with the framework.

Conclusions:

• The MGAC framework demonstrates capability in predicting flexible molecule crystal structures.
• Further refinement of the underlying potential is necessary for enhanced accuracy and ranking.
• The study highlights the ongoing need for robust computational tools in crystallography.