CRISPRi perturbation screens and eQTLs provide complementary and distinct insights into GWAS target genes

Abstract

Most genetic variants associated with human traits and diseases lie in noncoding regions of the genome 1, and a key challenge is determining which genes they affect 2,3. A common approach has been to leverage associations between natural genetic variation and gene expression to identify expression quantitative trait loci (eQTLs) in the population 4,5. At the same time, a novel approach uses pooled CRISPR interference (CRISPRi) perturbations of noncoding loci with single-cell transcriptome sequencing 6,7. Here, we systematically harmonized and compared the results from these approaches across hundreds of genomic regions associated with blood cell traits. We find that while the two approaches sometimes identify the same target genes, there are considerable differences that affect biological inferences made from the data. CRISPRi preferentially maps highly proximal, constraint-enriched genes, whereas eQTLs recover multiple, often distal targets. By benchmarking against 1,075 gold-standard CRE-gene pairs linked to blood traits, we show that the two approaches identify largely distinct targets; when combined, they achieve a balance between accuracy and completeness of gene discovery. Our results offer guidance for improved design of CRISPRi and eQTL studies and highlight their joint potential as a powerful toolkit for interpreting disease-associated loci.