CAREER: Interpretable machine learning deciphers single-cell multi-modal data for understanding cell-type functional genomics in complex brains

点击翻译按钮获取中文摘要

This award is funded in whole or in part under the American Rescue Plan Act of 2021 (Public Law 117-2).Brains are made up of billions of cells with different functions and have dramatically drawn research and public attention. However, underlying molecular mechanisms of brain cell functions are unclear. Recent advances in single-cell technologies enable measuring different characteristics (multi-modal data) of thousands of individual cells in complex brains, such as gene expression patterns, cell shapes, and behaviors. However, integrating such complex multi-modal data and interpreting molecular mechanisms from the data for brain cell functions remains challenging. This project will develop machine learning methods to build roadmaps linking multi-modal data of brain cells, revealing unseen data connections, insights into biological mechanisms, and improving prediction of cellular phenotypes and functions. The developed methods will be open-source and available for broadening community use. The project will also foster the integration of research and education through STEM programs, seminars, courses, online learning and provide publicly available materials. These activities will enhance participation and scientific understanding of minorities, underrepresented groups, and families with intellectual or neurodevelopmental disabilities, especially for machine learning in brain research.The project will deliver novel machine learning methods to predict cellular phenotypes and functions from multi-modal data of single cells and decipher cell-type functional genomics and gene regulation, a key molecular mechanism in brain cell functions. Aim 1 will develop a manifold learning method to align general single-cell multi-modalities (beyond multi-omics) and identify genes for predicting other modalities of brain cells (e.g., electrophysiology and morphology). Aim 2 will develop a comparative network analysis to reveal the relationships of multiple cell-type gene regulatory networks, revealing potential novel cell-type conserved and specific regulatory mechanisms. Aim 3 will develop a deep neural network model to prioritize “multi-modal networks” linking potentially causal genes and networks and other modal features for cellular phenotypes and functions. The results of this project can be found at https://daifengwanglab.org/.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.