Cell-type-specific immune genes seemed to form accurate links with nearby peaks accessible in the same cell types expressing the genes (Fig. 2l). We sought to systematically assess the accuracy of peak–gene links identified by examining a set of expression quantitative trait loci (eQTL) fine-mapped variants for whole blood, produced by the GTEx Consortium51. Of the 154,504 peaks in the PBMC dataset, only 3,598 contained a fine-mapped eQTL variant identified for whole blood by the GTEx Consortium and 2,054 of these peaks also had one or more peak–gene links identified. For the set of 2,054 peaks that overlapped a fine-mapped eQTL and were linked to a gene, the eQTL variant was associated with the same gene as the peak in 52.6% of cases, whereas 13.4% was expected by random chance. The systematic identification of putative regulatory targets for any open chromatin region in the genome using multimodal single-cell datasets has the potential to enable a more accurate assignment of trait- or disease-associated noncoding variants to a gene likely to be impacted by the variant.
