Designing molecules with certain desirable properties is a topic, also referred to as de novo molecular generation, of interest to many research fields. It can be seen as a combinatorial optimization problem in which the goal is to explore the chemical space in search of molecules for further study or synthesis. The estimated size of the chemical space of even drug-like molecules alone is so huge that is not possible to explore it all. We present an approach, implemented in a software tool called alvaBuilder, which constraints the search to molecules that can be composed of a set of fragments. Such fragments are identified by splitting the molecules contained in a given training data set. The exploration is then performed by assembling the aforementioned fragments to generate candidate molecules. Each candidate is associated to a score based on the desired properties defined by the researcher. By using Genetic Algorithms, alvaBuilder generates a final pool with the molecules that have the best score. The Genetic Algorithms take inspiration by Darwinian natural selection assuming that the best fitted members of a population of molecules survive and improvements can be achieved by mutating and recombining their genes, i.e. their molecular fragments. The purpose of this talk is to present how to perform de novo molecular generation using alvaBuilder with some practical sample cases to design molecules with certain phisicochemical properties.

De novo molecular design with genetic algorithms using alvaBuilder

Fall 2022 abstracts, posters, and presentations will launch in mid-September.

ACS Fall 2022

The official archive and repository for all ACS Meetings & Expositions abstracts and select posters or presentations.