An Aggregation Advisor for Ligand Discovery.

Area of Science:

• Medicinal Chemistry
• Biochemistry
• Computational Chemistry

Background:

• Colloidal aggregation of organic molecules is a primary cause of artifactual inhibition of proteins.
• Despite understanding, reliably predicting aggregation remains a challenge in drug discovery.
• Over 12,000 aggregating organic molecules identified, enabling precedent-based prediction.

Purpose of the Study:

• To develop and validate a predictive method for identifying potential organic molecule aggregators.
• To assess the prevalence of aggregator-like compounds in scientific literature and commercial databases.

Main Methods:

• Investigated a predictive approach using lipophilicity, affinity, and similarity to known aggregators.
• Utilized Tanimoto coefficients (Tc's) to quantify molecular similarity to known aggregators.
• Prospectively tested predictions experimentally.

Main Results:

• Five of seven new molecules with high similarity (Tc 0.95-0.99) to known aggregators showed aggregation.
• Significant aggregation observed in compounds with lower similarity scores (Tc 0.85-0.94).
• 5.1% of medicinal chemistry literature ligands (0.1-10 μM) are highly similar to known aggregators, compared to 0.73% of commercial compounds.

Conclusions:

• The developed method effectively predicts potential colloidal aggregators based on molecular properties and similarity.
• Aggregator-like compounds are enriched in the scientific literature, increasing over time.
• An open-access aggregator advisor database and tool are provided to aid researchers in distinguishing true hits from artifacts.