Page d'accueil

EVO-PERPHECT: Artifical Evolution on Natural Viral Genomes

Due to the excessive and inappropriate use of antibiotics, bacteria have developed resistance to them, which has necessitated the exploration of alternative treatments like phage therapy. Phages, which are viruses that can eliminate bacteria without harming beneficial bacteria, have shown potential in this regard. Nevertheless, two challenges impede the efficacy of phage therapy: host range and the relationship between bacteria and/or phages. To overcome the host range issue, Professor Pena and his team developed a Deep Learning model to anticipate potential interactions between phages and bacteria. Meanwhile, genetically engineered (GE) phages are proposed to address the second constraint. This project investigates the use of Evolutionary Algorithms (EA) to enhance phages via a distinctive gene-editing approach. The study encompasses reviewing crucial molecular biology and EA topics, data analysis, hypothesis testing, and result evaluation and discussion.

This study highlights the importance of various factors, such as fitness functions, operations, and hyperparameters, on phage evolution. It reveals that the optimal balance between host range and similarity is crucial for desired outcomes in phage-related applications. Simulations incorporating both crossovers and mutations prove more effective, and despite the consistent convergence of fitness values, the algorithm might not be efficiently exploring the search space for better solutions. Genetic Algorithms demonstrate exceptional performance in host range optimization, emphasizing the need for accurately representing the relationship between host range and similarity. Overall, this research underscores the complexity of the evolutionary process and the importance of carefully selecting fitness functions, operations, and hyperparameters in phage evolution simulations.