AI for Predictive Genomics: From Sequence to Disease

Graduate / Professional

This course examines the frontiers of computational genomics in the era of artificial intelligence, focusing on how modern AI/ML models are reshaping our ability to map sequence to regulation to cell state to disease. Rather than a broad survey, the course is organized around core problem classes—including biological sequence modeling, single-cell and spatial data integration, variant-to-function mapping, gene regulatory networks, perturbation modeling, and disease gene discovery. The course emphasizes a unifying view of genomics as a predictive problem linking DNA sequence to cellular and disease phenotypes through AI-driven models. Students will critically analyze state-of-the-art methods (e.g., foundation models, deep generative models, and large language models) and learn how they are used to solve real biological problems. Emphasis is placed on connecting methodological advances to biological questions, understanding model assumptions and limitations, and identifying open challenges. Through literature presentations and a project-driven format, students will design, implement, and evaluate AI/ML approaches on real genomics problems. The course also features invited lectures from researchers developing influential methods, providing direct exposure to ongoing work at the cutting edge. This is an advanced course in computational biology and does not provide a comprehensive introduction to basic concepts or methodologies. Students are expected to have completed entry-level coursework in probability and statistics, artificial intelligence (AI) or machine learning (ML), and genetics or genomics. Students should also be comfortable programming in Python or a similar language, as the final project involves implementing AI/ML methods to address real-world problems in computational genomics. Please contact the instructor if you have any questions regarding these prerequisites. HTBA