DeepCOMO: from structure-activity relationship diagnostics to generative molecular design using the compound...

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Drug Discovery

Background:

• The Compound Optimization Monitor (COMO) is a diagnostic tool for evaluating lead optimization progress.
• It assesses chemical saturation within analog series using virtual populations.
• COMO bridges optimization diagnostics and compound design.

Purpose of the Study:

• To introduce DeepCOMO, a deep learning extension of COMO for generative molecular design.
• To illustrate DeepCOMO's capabilities in diagnosing analog series and guiding optimization.
• To evaluate different analog design strategies and prioritize virtual candidates.

Main Methods:

• Deep learning extension of the COMO approach, termed DeepCOMO.
• Application of DeepCOMO on exemplary analog series.
• Diagnostic assessment of chemical saturation and structure-activity relationship (SAR) progression.

Main Results:

• DeepCOMO effectively diagnoses chemical saturation and SAR progression.
• The tool evaluates various analog design strategies.
• Prioritization of virtual candidates for optimization based on development stage.

Conclusions:

• DeepCOMO enhances generative molecular design capabilities.
• The approach provides comprehensive diagnostics for analog series development.
• DeepCOMO aids in efficient and informed compound optimization efforts.