The Molecule Evoluator. An Interactive Evolutionary Algorithm for the Design of Drug-Like Molecules

Abstract

To help chemists design new drugs, we created a tool that uses interactive evolution to design drug molecules, the "Molecule Evoluator". In contrast to most other evolutionary de novo design programs, the molecule representation and the set of mutations enable it to both search the chemical space of all drug like molecules extensively and to fine-tune molecular structures to the problem at hand. Additionally, we use interaction with the user as a fitness function, which is new in evolutionary algorithms in drug design. This interactivity allows the Molecule Evoluator to use the domain knowledge of the chemist to estimate the ease of synthesis and the biological activity of the compound. This knowledge can guide the optimization process and thereby improve its results. Chemists of our department using the Molecule Evoluator were able to find six novel and synthesizable druglike core structures, indicating that the Molecule Evoluator can be used as a tool to enhance the chemist's creativity.

Combining Aggregation with Pareto Optimization: A Case Study in Evolutionary Molecular Design

“…This problem has a large number of criteria which can be partitioned into fuzzy or soft constraints, such as the solubility of a candidate molecule, and typically a small set of objectives which in our case are the maximization of predicted activities on certain targeted cells. This study extends the work presented in [8,13].…”

“…This can be attributed to an increasing quality of methods to predict the biological and pharmacological properties of molecules, as well as increased knowledge in the evolutionary algorithms field on how to deal with the complex search spaces induced by the graph-like structures of molecules. Successful recent approaches are described in [19,8,16,13].…”

Chemical space as a source for new drugs

“…78 EVOLUATOR has been used to identify compound 8 as an inhibitor that is active on both the a1-and a1-adrenergic receptors and shows a displacement of >50% at a concentration of 10 mM in the radioligand binding assay. 79 Skelgen has been used to discover estrogen inhibitors. From the 17 synthesized structures, 5 show inhibition in mM range, such as 9 (IC 50 ¼ 0.34 mM).…”

Analyzing molecular landscapes using random walks and information theory

“…Given the data from random walks the following analysis methods can be used: (1) Correlation Length Analysis which reveals the validity of the continuity assumption; (2) Information Complexity which reveals the structural diversity of the search landscape; (3) Multimodality measures which estimate the frequency of local optima for different neighbourhood radii, and finally (4) Neutrality measures which account for the size distribution of plateaus in the landscape. Each measure indicates difficulties for optimization routines encountered when optimizing the objective function.We apply random-walk based landscape analysis for two search spaces in the context of de-novo drug design: Firstly, we study the properties of the search space induced by the mutation operators of the Molecule Evoluator™ [2] with an activity model as an objective function. Secondly, we study the properties of a peptide design problem using the software MOE.…”

“…We apply random-walk based landscape analysis for two search spaces in the context of de-novo drug design: Firstly, we study the properties of the search space induced by the mutation operators of the Molecule Evoluator™ [2] with an activity model as an objective function. Secondly, we study the properties of a peptide design problem using the software MOE.…”