Multi-objective protein encoding: Redefinition of the problem, new problem-aware operators, and approach based on Variable Neighborhood Search

Abstract

Synthetic biology is currently involved in improving the expression levels of proteins. A preeminent strategy herein is to encode multiple copies of the same amino acid sequence (protein) and integrate these copies into the genome of an organism. However, this is a challenging task. First, the different protein-coding sequences (CDSs) should be as different as possible to avoid homologous recombination. The second challenge lies in the fact that some synonymous codons are better adapted than others. Therefore, the best adapted synonymous codons should be employed if possible. These two challenges are conflicting and multi-objective optimization has to be consequently used. For this purpose, we design and implement a Multi-Objective Variable Neighborhood Search (MOVNS) algorithm with new problem-aware mutation operators. Following the related work, we initially define the multi-objective problem based on three optimization objectives. However, after the first experiments, we realize that one of the objectives can be discarded, simplifying the problem without degrading the solutions’ quality. The obtained results confirm the attainment of statistically significant improvements in comparison to related works.