MOFSocialNet: Exploiting Metal-Organic Framework Relationships via Social Network Analysis.

Area of Science:

• Materials Science
• Computational Chemistry
• Data Science

Background:

• The rapid growth in metal-organic frameworks (MOFs) and their applications necessitates advanced data analysis tools.
• Manual sorting of over 100,000 characterized MOFs is infeasible.
• Data science, particularly graph theory and machine learning, offers solutions for navigating the MOF chemical space.

Purpose of the Study:

• To apply social network analysis (SNA) to the chemical space of MOFs for the first time.
• To introduce MOFSocialNet, a novel social network based on MOF geometrical descriptors.
• To demonstrate the utility of SNA for MOF community detection, representative MOF identification, and property prediction.

Main Methods:

• Utilized graph theory, specifically social network analysis (SNA).
• Developed MOFSocialNet using geometrical descriptors from the CoRE-MOFs database.
• Applied community detection and graph node centrality algorithms.

Main Results:

• MOFSocialNet successfully identified communities of structurally similar MOFs.
• The tool can pinpoint the most representative MOFs within identified communities.
• SNA analysis of MOFSocialNet improved the prediction accuracy of MOF properties, such as gas storage, compared to conventional machine learning.

Conclusions:

• Social network analysis is a powerful tool for understanding the MOF chemical space.
• MOFSocialNet provides a novel framework for MOF discovery and property prediction.
• This approach enhances the selection of MOFs for specific applications, particularly for gas storage.