LEADD: Lamarckian evolutionary algorithm for de novo drug design

Area of Science:

• Computational chemistry
• cheminformatics
• Drug discovery

Background:

• De novo molecular design aims to identify molecules optimizing specific properties.
• A key challenge is the generation of synthetically unfeasible molecules.
• Existing methods often struggle to balance optimization with synthetic accessibility.

Purpose of the Study:

• To present a novel Lamarckian evolutionary algorithm for de novo drug design (LEADD).
• To enhance the synthetic accessibility of designed molecules while maintaining optimization power.
• To improve computational efficiency in molecular design.

Main Methods:

• LEADD represents molecules as graphs of molecular fragments.
• It utilizes knowledge-based atom type compatibility rules to limit bond formation.
• A Lamarckian evolutionary mechanism adapts molecular reproductive behavior for efficient chemical space sampling.

Main Results:

• LEADD was compared against standard virtual screening and a comparable evolutionary algorithm.
• The algorithm demonstrated improved efficiency in identifying fitter molecules.
• Molecules designed by LEADD are predicted to have higher synthetic accessibility.

Conclusions:

• LEADD offers a balanced approach to de novo molecular design, optimizing for properties and synthetic feasibility.
• The algorithm represents a significant advancement in generating practical drug candidates.
• LEADD enhances computational efficiency and chemical space exploration in molecular design.