Home // International Journal On Advances in Life Sciences, volume 6, numbers 3 and 4, 2014 // View article
Authors:
Susanne Rosenthal
Markus Borschbach
Keywords: multi-objective biochemical optimization, population size, landscape analysis, multi-parent recombination.
Abstract:
The main task in the drug design process is the prediction of the peptide structure and the bioactivity with the focus on simultaneously optimization of molecular peptide features. The synthesis and laboratory screening are the conventional but cost-intensive steps for optimization. Multi-objective genetic algorithms provide a range of methods for an efficient design of drug peptides. A customized NSGA-II has been especially evolved for biochemical optimization with the focus on producing a great number of very different high quality peptides within a very low number of generations e.g., under 20, termed early convergence. The focus of this work are an insight into the impact of the interdependence between the selection procedure and the population size, the empirical verification of the early convergence behavior within a limited range of population size and the influence of multi-parent recombination on the algorithm performance. These purposes are exemplary investigated on two different dimensional biochemical optimization problems, which are concrete, but as generic as possible. A landscape analysis is performed to gain an insight into the characteristic features and difficulties of the multi-objective optimization problems. The performance is assessed on the basis of a convergence indicator especially evolved for our preference of comparing the convergence behavior of populations with different sizes.
Pages: 310 to 324
Copyright: Copyright (c) to authors, 2014. Used with permission.
Publication date: December 30, 2014
Published in: journal
ISSN: 1942-2660