Abstract

Drug discovery is inherently a multicriteria optimization problem. In the first instance, it involves a tremendously large chemical space where each compound can be characterized by multiple molecular and biological properties. Modern computational approaches try to efficiently explore chemical space in search of molecules with the desired combination of properties. For example, Pareto optimizers identify a so called “Pareto front”, a set of nondominated solutions. From qualitative perspective all solutions on the front are potentially equally desirable, each expressing a tradeoff between the goals. However, often there is a need to weight the objectives differently, depending on their perceived importance. To address this, we have recently implemented a new Multicriteria Decision Aiding/Analysis (MCDA) method as a part of our AI powered Drug Design (AIDDTM) technology initiative [1]. This allows the user to differently weight various objective functions, which, in turn, efficiently directs the generative chemistry process towards the desired areas in chemical space.

By Rafał A. Bachorz