Structural diversity of biologically interesting datasets: a scaffold analysis approach

Area of Science:

• Medicinal Chemistry
• Drug Discovery
• Computational Chemistry

Background:

• Publicly available metabolome and natural product datasets are valuable for drug design.
• Characterizing scaffold diversity in biologically relevant datasets is crucial for lead library development.
• Comparative analysis of current drug leads against biological datasets is needed.

Purpose of the Study:

• To compare scaffold diversity and physicochemical properties of current drug leads with metabolite and natural product datasets.
• To identify opportunities for improving lead libraries by incorporating underutilized scaffolds.
• To inform drug design strategies by understanding the chemical space of biologically relevant compounds.

Main Methods:

• Comparative analysis of scaffold distribution in drug, lead, metabolite, and natural product datasets.
• Analysis of physicochemical properties including molecular polar surface area, solubility, number of rings, and rotatable bonds.
• Evaluation against established drug-likeness rules, such as Lipinski's rule.

Main Results:

• Metabolite scaffolds are enriched in drug datasets (42%) compared to lead libraries (23%).
• Lead datasets share only a small fraction (5%) of the natural product scaffold space.
• Metabolites exhibit high polarity and solubility but low ring counts; natural products have high ring and rotatable bond counts.

Conclusions:

• Current lead libraries underutilize the scaffold diversity of metabolites and natural products.
• Incorporating metabolite and natural product scaffolds can enhance lead library design for targeted therapies.
• While beneficial, the limited chemical space distribution of metabolites may constrain their library design applications.