IV Jornadas Latinoamericanas de Bacteriófagos

This study presents a proof-of-concept for enhancing protein-protein interactions using a single-objective elitist evolutionary algorithm. The research focuses on a de novo structural complex formed by the bacteriophage Rs551's Receptor Binding Protein (RBP) and the R. solanacearum bacterial pilus. The primary objective was to minimize the interaction energy within the defined 38-position interface zone to design a more robust molecular interaction. To achieve this, a genetic algorithm (GA) was integrated with the Rosetta Macromolecular Modeling suite. The algorithm employs a specialized mutation operator that identifies stabilizing and destabilizing residues and probabilistically selects one for replacement, guided by a Multiple Sequence Alignment (MSA) probability matrix that reflects naturally occurring substitutions.

The experimental results demonstrate the effectiveness of this computational approach. Over 16 independent runs and 20 generations, the algorithm consistently found improved designs. A clear convergence was observed, with the average fitness value (interaction energy) decreasing from an initial average of 15.78 to -2.77 in the final generation, indicating a significant enhancement in binding affinity. Analysis of the final 400 solutions revealed specific mutation patterns, identifying key amino acid positions and substitutions that contribute to lower energy states. Interface quality metrics further confirmed that the evolved protein designs had better energy scores without significantly altering the interface's structural packing. This work successfully validates the evolutionary algorithm as a tool for automating the search of optimal solutions in protein design, effectively highlighting energetically favorable mutations. As future work, extensive validation and benchmarks must be incorporated, along with transforming the genetic algorithm from single objective to multi-objective.