Overview

• Northwestern University researchers created the Transcriptome-Wide conditional Variational auto-Encoder (TWAVE), a generative AI framework that amplifies sparse gene expression data to reveal causal gene sets.
• TWAVE simulates both healthy and diseased cellular states to pinpoint expression changes that trigger complex traits.
• Validation across diabetes, cancer and asthma showed TWAVE uncovered genes missed by conventional genome-wide association studies.
• Analysis revealed that different combinations of genes can underlie the same disease in different patients, suggesting opportunities for personalized therapies.
• The study appeared in the Proceedings of the National Academy of Sciences with support from the National Cancer Institute, National Science Foundation and Simons Foundation.