Toward an efficient approach to identify molecular scaffolds possessing selective or promiscuous compounds

Area of Science:

• Medicinal Chemistry
• Computational Drug Discovery
• Chemical Biology

Background:

• Recurrent chemical scaffolds are fundamental in medicinal chemistry and drug design.
• Understanding scaffold-driven selectivity and promiscuity is crucial for both traditional and poly-pharmacological approaches.
• A comprehensive scaffold analysis of a large, publicly available dataset with complete screening profiles was previously unreported.

Purpose of the Study:

• To analyze a unique dataset of over 15,000 compounds screened against 100 sequence-unrelated proteins.
• To identify specific chemical scaffolds that contribute to either promiscuous or selective binding.
• To develop a generalizable computational approach for scaffold analysis in drug discovery.

Main Methods:

• Utilized the Molecular Equivalence Index tool to identify Bemis-Murcko frameworks with at least five compounds.
• Employed Scaffold Hunter to construct a hierarchical scaffold tree.
• Annotated the scaffold tree with protein-binding data to link scaffolds to activity profiles.

Main Results:

• Successfully identified predominantly highly specific compounds, influenced by dataset constraints.
• Demonstrated the identification of scaffolds driving selectivity and promiscuity.
• Applied the methodology to a separate dataset of 1497 small molecules screened against 172 protein kinases.

Conclusions:

• The developed scaffold analysis protocol is effective for identifying structure-activity relationships.
• The approach provides valuable insights for optimizing compound selectivity and promiscuity in drug design.
• This generalizable method can be applied to various datasets and activity readouts for broader drug discovery applications.