Automatic Identification of Analogue Series from Large Compound Data Sets: Methods and Applications

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Drug Discovery

Background:

• Analogue series are vital in drug discovery, particularly during lead optimization.
• Identifying analogue series in large databases without predefined core structures presents a significant challenge.

Purpose of the Study:

• To review common and recent methodological developments for automatically identifying analogue series in large chemical libraries.
• To provide an overview of techniques for extracting molecular analogues based on core structures and modifications.

Main Methods:

• Review of scaffold extraction methods, including rule-based approaches like Bemis-Murcko scaffolds.
• Discussion of matched molecular pair (MMP) and matched molecular series (MMS) concepts for identifying similar compounds.
• Exploration of hierarchical scaffold decomposition and systematic molecule fragmentation for analogue series extraction.

Main Results:

• Evolution from rule-based scaffold extraction to MMP/MMS approaches.
• Development of algorithms for single-site modifications and multi-substitution site analogue series.
• Enabling efficient analysis of large chemical datasets (hundreds of thousands to millions of compounds).

Conclusions:

• Automated analogue series identification methods have significantly advanced drug discovery analysis.
• These techniques facilitate the representation of chemical series, such as R-group tables.
• Methodological developments continue to emerge, supporting large-scale compound library analysis.