Kolmogorov-Smirnov statistic and its application in library design

Area of Science:

• Medicinal Chemistry
• Computational Chemistry
• Drug Discovery

Background:

• Traditional combinatorial chemistry focuses on lead generation and molecular diversity, often neglecting drug likeness.
• Quantifying and incorporating drug likeness into experimental design has been a recent focus, but conventional methods have limitations.

Purpose of the Study:

• To propose a new paradigm for experimental design using multiobjective optimization to enhance drug likeness.
• To integrate traditional design objectives with secondary criteria for pharmacologically relevant chemical space exploration.

Main Methods:

• Utilized multiobjective optimization principles for experimental design.
• Incorporated a novel Kolmogorov-Smirnov statistic-based criterion to control drug likeness properties.
• Applied the method to design an exploratory library optimizing five parameters simultaneously.

Main Results:

• Demonstrated limitations of conventional methodologies in drug likeness assessment.
• Developed a robust and extensible method for balancing conflicting design objectives.
• Successfully designed a diverse library with drug-like molecular weight and logP profiles, suitable for array synthesis.

Conclusions:

• The proposed multiobjective optimization approach offers a powerful and flexible strategy for designing drug-like chemical libraries.
• This method enables medicinal chemists to achieve optimal compromises between multiple, often competing, design criteria.
• The approach facilitates the creation of libraries that are both diverse and possess properties aligned with known drugs, improving efficiency in drug discovery.