Abstract
Biological phenotypes arise from the degrees to which genes are expressed, but the lack of tools to precisely control gene expression limits our ability to evaluate the underlying expression-phenotype relationships. Here, we describe a readily implementable approach to titrate expression of human genes using series of systematically compromised sgRNAs and CRISPR interference. We empirically characterize the activities of compromised sgRNAs using large-scale measurements across multiple cell models and derive the rules governing sgRNA activity using deep learning, enabling construction of a compact sgRNA library to titrate expression of ∼2,400 genes involved in central cell biology and a genome-wide in silico library. Staging cells along a continuum of gene expression levels combined with rich single-cell RNA-seq readout reveals gene-specific expression-phenotype relationships with expression level-specific responses. Our work provides a general tool to control gene expression, with applications ranging from tuning biochemical pathways to identifying suppressors for diseases of dysregulated gene expression.
