Modeling TMPRSS2/SLC6A20 Expression in Lung Cell Types

We used MuSiC (v.0.1.1) (Wang et al., 2019 (link)) to estimate the proportions of lung cell types with >500 cells from our scRNA-seq dataset in lung bulk RNA-seq samples from the GTEx v8 release (GTEx Consortium, 2020 (link)). We combined cell-type labels for capillary (distal and proximal), macrophages (M1 and M2), matrix fibroblasts (1 and 2), and NK/T cells. We modeled the relationship between TMM-normalized TMPRSS2 or SLC6A20 expression as a function of the interaction between genotype and cell-type proportion, while considering the covariates used in the original GTEx data including sex, sequencing platform, PCR, five genotype PCs, and 59 inferred PCs from the expression data. From the original inferred PCs, we excluded inferred PC one because it was highly correlated with AT2 cell-type proportion (Spearman ρ = 0.67). Including additional covariates in the model such as age, body-mass index or smoking status did not have meaningful impact on the results.