Machine Learning Clustering of Water-Water Interactions in the Cambridge Structural Database.

Area of Science:

• Crystallography
• Computational Chemistry
• Machine Learning

Background:

• Water molecules play a crucial role in crystal structures.
• Understanding water-water interactions is key to comprehending molecular behavior and crystal packing.
• Existing methods for analyzing these interactions can be enhanced with advanced computational techniques.

Purpose of the Study:

• To apply clustering machine learning techniques for analyzing water-water contacts in crystal structures.
• To identify and categorize different types of water-water interactions based on their geometric parameters.
• To demonstrate the value of machine learning in uncovering insights into molecular interactions.

Main Methods:

• Utilized clustering machine learning algorithms.
• Analyzed crystal structures from the Cambridge Structural Database.
• Grouped water-water contacts based on interaction energies.
• Defined geometrical parameters for identified contact groups.

Main Results:

• Successfully identified similar groups of water-water contacts using machine learning.
• Defined characteristic geometrical parameters for these distinct contact groups.
• Visual examination of clustering results provided valuable insights into interaction diversity.

Conclusions:

• Clustering machine learning is effective for analyzing water-water contacts in crystal structures.
• This approach enhances our understanding of the diverse spectrum of water molecule interactions.
• Integrating clustering methods into visualization software can facilitate the discovery of novel interactions.