An alphabetic code based atomic level molecular similarity search in databases.

Area of Science:

• Bioinformatics
• Chemo-informatics
• Computational Chemistry

Background:

• Molecular similarity and diversity studies are crucial for drug design.
• Large chemical compound datasets necessitate efficient search methodologies.
• Existing methods may lack optimal computational efficiency for large-scale searches.

Purpose of the Study:

• To develop a novel alphabetic algorithm for efficient molecular similarity searching.
• To enable rapid retrieval of active compounds from large chemical databases.
• To improve computational efficiency in drug discovery processes.

Main Methods:

• An alphabetic algorithm based on atom-atom bonding preference was developed.
• 170 cyclin-dependent kinase 2 inhibitors were represented as pre-defined alphabet strings.
• Protein sequence alignment tools were adapted for searching these strings.
• Area under the receiver operating characteristic (ROC) curve was used for performance evaluation.

Main Results:

• The algorithm successfully extracted common patterns from inhibitor datasets.
• An average retrieval rate of approximately 60% was achieved in cross-validation.
• Performance was validated using both in-house and public (DUD) datasets.
• The method demonstrated effective retrieval of similar active compounds.

Conclusions:

• The developed alphabetic algorithm offers an efficient approach for molecular similarity searching.
• This method enhances the ability to discover similar active compounds in large databases.
• The approach holds significant potential for accelerating drug design and discovery pipelines.