The research focuses on the generative design of novel bacteriophages using genome language models to create viable phages with specific host tropism and functional properties. This involves using these models to generate whole-genome sequences and experimentally identify phages with diverse fitness profiles and novel structural features. Additionally, the group explores using deep learning architectures like LSTMs to modify existing phage genomes to improve characteristics such as host range, and developing frameworks to quantify evolutionary novelty and design efficiency in these generative models.