Machine Learning Augmented Discovery of AAV Capsids for Cell Type Specific Access into Human Neurons and Glia

Project Summary The human brain contains an astonishing diversity of neuronal and glial cell types distributed across dozens of functional areas. Single cell genomics resources generated by the BRAIN Initiative Cell Census have offered a wealth of reference knowledge about this diversity, including cell types present in human and non-human primate

brain. However, we currently lack scalable technologies to genetically access these specific cell types, which will be necessary to advance our understanding of their function. By leveraging our expertise in viral engineering, single cell analysis, machine learning, and non-human and human primate neuroscience, we propose to develop a scalable toolkit for accessing distinct cell types in the

human brain. Specifically, our U01 application will develop novel adeno-associated viral (AAV) vectors for cell type-specific gene delivery using machine learning augmented engineering of capsid libraries. In this pilot grant, we seek to demonstrate the potential of our approach to support the development of a comprehensive

Armamentarium for brain cell types that would enable functional and translational studies across a broad range of applications. In particular, we will apply this methodology to develop AAV vectors for cell types in the human forebrain, including the hippocampus. Our tools will be openly disseminated after validation for specificity using molecular

and physiological metrics, and mapped onto existing single cell genomics resources from the same brain regions generated by the BRAIN Initiative Cell Census projects. If successful, our project will establish a scalable infrastructure for addressing and functionally investigating in principle any tissue or cell type in the non-human or human primate brain.