ADMET-Driven Chemical Navigability Rules for Early-Stage Virtual Compound Prioritization

Area of Science:

• Medicinal Chemistry
• Drug Discovery
• Computational Chemistry

Background:

• Clinical drug development faces high attrition rates due to poor pharmacokinetic and toxicological (ADMET) properties.
• Early assessment of ADMET properties is crucial in Structure-Based Drug Design (SBDD).

Purpose of the Study:

• To develop empirically derived chemical navigability guidelines for early-stage ADMET assessment.
• To integrate these guidelines into a rapid compound assessment framework.
• To evaluate the effectiveness of an ADMET-first strategy in drug discovery.

Main Methods:

• Analysis of ADMET properties predicted by admetSAR 3.0 for approved drugs from DrugBank.
• Development of a color-coded visualization framework for compound assessment.
• Evaluation using an independent library of KEAP1/NRF2 modulators from ChEMBL, assessing experimental biological activity (pChEMBL).

Main Results:

• Enrichment analysis showed that ranking compounds by ADMET-score identified active compounds earlier than random selection (EF=1.29 in top 5%).
• Approximately 80% of active compounds were recovered within a small fraction of the chemical space.
• A freely accessible database of over two million ADMET-annotated compounds, prefiltered by the guidelines, is provided.

Conclusions:

• The proposed chemical navigability guidelines, based on ADMET properties, enhance the efficiency of drug discovery.
• An ADMET-first approach significantly improves the early identification of viable drug candidates.
• The developed framework and database offer valuable resources for Structure-Based Drug Design and lead optimization.