Fig. 6: Fine-mapping OAS1 locus using three different model systems.

a, Schematic of in vivo system, COVID-19 study of PBMCs and nasal brushings to confirm the splicing QTL association in the OAS1 gene. b, Locus zoom plots show the COVID-19 GWAS (COVID-19 versus population) association Bayes factors as well as eQTL associations of the OAS1 gene in fibroblasts and PBMCs. c, UMAP shows the expression levels of the OAS1 gene in fibroblasts. UMAP coordinates are identical to Fig. 2a. d, UMAP shows the eQTL effect size of the OAS1 gene at rs10774671G>A. UMAP coordinates are identical to Fig. 2a. e, UMAP shows OAS1 expression level in PBMCs. UMAP coordinates are identical to Extended Data Fig. 7a. f, UMAP shows the eQTL effect size of the OAS1 gene at rs10774671G>A in PBMCs. UMAP coordinates are identical to Extended Data Fig. 7a. g, Sequencing coverage depth around the splicing variant rs10774671G>A, which creates three different isoforms, two of which are not annotated in Ensembl 90. scRNA-seq reads in fibroblasts were aggregated and stratified by the three different genotype groups (GG, reference homozygote; GA, heterozygote; AA, alternative homozygote). h, The 10x RNA-seq coverage depth of epithelial cells around OAS1 3′ end in nasal brushing samples taken from 33 COVID-19-positive adult patients was stratified and aggregated by the three genotype groups of rs10774671G>A as demonstrated in g. UCL, University College London.