Project Summary Rapidly evolving proteins constitute a significant portion of a bacterial proteome. These proteins diverge substantially from their homologs, frequently change functions and locations of active sites, and therefore are more challenging to study. Fast-evolving proteins are nevertheless as crucial as conserved ones for our fundamental understanding of molecular evolution and biomedical applications: pathogenicity factors in bacterial pathogens typically undergo fast evolution due to arms races between hosts and pathogens. Revolutionary advancements in computational protein science now facilitate the efficient investigation of fast- evolving proteins. Advanced artificial intelligence methods such as AlphaFold have produced accurate models of protein 3D structures, which could revolutionize bioinformatics approaches. Now, instead of relying on sequence-based homology searches, which frequently fail to find relatives of fast-evolving proteins, we can use similarity in 3D structures, which tend to be more conserved evolutionarily. Additionally, methods for predicting protein-protein interactions (PPIs) are nearing the accuracy of high-throughput experiments, offering another approach to gain functional insights to a protein by finding a well-studied interaction partner. The time is ideal to capitalize on these developments and tackle challenging problems that were previously impossible to address. In this proposal, I intend to study fast-evolving proteins using state-of-the-art computational methods, followed by experimental validation. First, I have developed methods to recognize domains from AlphaFold models and assign them to an evolutionary context for functional inference. I plan to enhance these methods using deep learning techniques and develop additional methods to predict functional categories. While existing tools are primarily optimized for conserved protein families with deep sequence alignments and extensive experimental data, my methods will cater specifically to fast-evolving proteins. Subsequently, I will apply these tools and my expertise in comparative genomics and PPI modeling to study the fast-evolving and pathogenicity-associated proteins encoded by the pan-genomes of Vibrio parahaemolyticus (Vpara) strains isolated from human patients. By determining their evolutionary origins, predicting their interacting partners, and inferring their functions, I aim to identify a set of uncharacterized and fast-evolving candidate pathogenicity factors (CPFs). Finally, I will select several promising CPFs and experimentally validate their secretion and interacting partners using medium- throughput experiments. This preliminary functional characterization will be the initial step toward elucidating the mechanisms of these novel PFs and will pave the way for future discoveries in my lab and in the field. This project will allow me to undergo rigorous and multidisciplinary training in cutting-edge computational methods and experimen...