Subproject: Evolutionary Algorithms for Drug Design

Departments and research groups of the faculties of biology, organic biology/bio-chemistry, physical chemistry, physics and bioinformatics participate in the interdisciplinary research project "drug design by parallel multi parameter analysis of molecular complexes". It is the aim of the project to get potential drugs more quickly by carrying through an extensive quantitative analysis of the molecular interaction of a pathologically relevant molecular complex, combined with activity predictions through bioinformatics. The subproject assigned to bioinformatics specializes in the use of evolutionary algorithms.

Applications of evolutionary algorithms (especially genetic algorithms) for drug design have not been published until the 1990s. In doing so, classical codings were used for the individuals. In this subproject special codings for polypeptides or proteins based on the alphabet of amino acids are designed and a genetic algorithm for these codings is implemented.

In order to verify that the developed genetic algorithm is able to optimize the peptides due to given criteria, different optimization problems are implemented. In case of the optimization problem outlined in the figure a given target peptide (optimum) shall be found. An alignment between the individual of the population to be evaluated and the target peptide is used to determine the fitness (alignment score) of that individual.



Example of a fitness function based on alignment

During the project, the GA shall be used to suggest new potential peptides. These peptides will be tested  by our project partners as to their efficiency and the results will flow back into the GA.