A large-scale single cell map of primary and conditional regulatory variation in the human brain

Abstract

Most disease-associated genetic variants reside in noncoding regions and are thought to influence risk by altering gene regulation. However, linking these variants to target genes and mechanisms remains challenging because regulatory effects can be cell type specific and may involve multiple independent signals at the same locus. Although eQTL studies have identified many regulatory variants, many GWAS loci do not colocalize with primary eQTLs, suggesting that additional regulatory signals remain unresolved. Here, we generated a large-scale single cell genomic resource to map primary and conditional regulatory variation in the human brain. We integrated single nucleus RNA-seq data from 956 postmortem brains, comprising 4.8 million nuclei across seven major brain cell types, with matched whole genome sequencing data. Using data from a subset of 763 individuals of European ancestry, we identified single cell eQTLs (sc-eQTLs) for 32,189 eGenes across cell types and uncovered 6,174 additional conditionally independent sc-eQTL signals through stepwise conditional analysis. Conditional sc-eQTLs were more cell type-specific than primary sc-eQTLs and showed distinct regulatory architectures, including greater distances from TSSs and reduced enrichment in annotated promoters and enhancers. Genes with conditional sc-eQTLs also showed lower genic constraint in most cell types, suggesting that genes with secondary regulatory effects are shaped by different selective pressures. Importantly, several schizophrenia and Alzheimers disease GWAS loci colocalized more strongly with conditional than primary sc-eQTLs, demonstrating that secondary regulatory signals can reveal disease-relevant mechanisms missed by primary sc-eQTL analyses alone. Together, these findings show that conditional and cell type-specific regulatory variation represents a substantial component of the genetic architecture of brain disease.