A new study published in Human Gene Therapy introduces a machine learning (ML) model that predicts the fitness of adeno-associated virus (AAV) capsids for gene therapy. This in silico approach is designed to streamline the process of developing AAV capsids, which is crucial for reducing manufacturing costs and making gene therapies more affordable for patients. Adeno-associated viruses (AAVs) are widely used as vectors in gene therapy due to their ability to deliver therapeutic genes to target cells. However, the development of AAV capsids with high yields, or fitness, is a complex and resource-intensive process. Traditionally, this has involved laborious in vitro experiments, which are time-consuming and expensive. Christian Mueller and his team from Sanofi have developed a state-of-the-art ML model, known as the capsid-protein language model (CAP-PLM), to address this challenge. The CAP-PLM predicts the fitness of AAV2 capsid mutants based on their amino acid sequences. By integrating a protein language model (PLM) with classical ML techniques, the researchers achieved a high level of prediction accuracy, with a Pearson correlation of 0.818. This robust performance was validated on completely independent datasets, demonstrating the model's reliability and generalizability, even for capsids with multiple mutations. The significance of this study lies in its potential to revolutionize the field of capsid engineering. Directed evolution and rational design, the traditional methods for enhancing AAV capsid fitness, are often limited by their trial-and-error nature and the high costs associated with extensive experimentation. The CAP-PLM, on the other hand, offers a more systematic, comprehensive, and cost-effective solution. Thomas Gallagher, Ph.D., Managing Editor of Human Gene Therapy from the University of Massachusetts Chan Medical School, praised the study. "The emergence of AI-based approaches in capsid engineering is an exciting development that can be more systematic, comprehensive, and cost-effective than traditional methods," he noted. "The work by Mueller and his team represents a significant step forward in developing AI tools for gene therapy." In summary, the CAP-PLM developed by Mueller and his colleagues at Sanofi is a groundbreaking ML model that can accurately predict the fitness of AAV2 capsid mutants. This advancement not only promises to reduce the time and cost involved in gene therapy development but also paves the way for more efficient and accessible treatments for a wide range of genetic disorders.