Derivatization Design of Synthetically Accessible Space for Optimization: In Silico Synthesis vs Deep Generative...

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Drug Discovery

Background:

• Molecular design is critical for lead optimization, impacting project timelines and costs.
• Traditional design cycles face challenges in speed, quality, and creativity.
• Deep learning-based generative design is emerging for de novo molecule design.

Purpose of the Study:

• To introduce and evaluate a novel de novo design technology called "derivatization design".
• To demonstrate the application of AI-assisted forward in silico synthesis for generating lead analogues and scaffold variations.
• To highlight the integration of synthetic feasibility, reagent availability, and cost data into the design process.

Main Methods:

• Development of a "derivatization design" technology utilizing artificial intelligence.
• Application of AI-assisted forward in silico synthesis for generating near neighbor analogues and scaffold variations.
• Inclusion of synthetic feasibility, reagent availability, and cost assessment within the automated design process.

Main Results:

• Derivatization design successfully generates project-relevant analogue sets with integrated synthetic data.
• The methodology provides early assessment of synthetic feasibility, reagent cost, and availability.
• A case study on DDR1 inhibitors demonstrated the effectiveness of derivatization design compared to deep generative design.

Conclusions:

• Derivatization design accelerates lead optimization by providing rapid, feasible, and cost-aware molecular designs.
• The integrated synthetic assessment facilitates early decision-making and reduces overall project timelines.
• This AI-driven approach offers a powerful alternative for de novo molecular design in drug discovery.