To calculate cell type specificity, we adapted the cell type specificity code from github.com/jbryois/scRNA_disease/blob/master/Code_Paper/Code_Zeisel/get_Zeisel_Lvl4_input.md (ref. 75) for these additional datasets. Briefly, we converted Seurat objects77 into SingleCellExperiment (v.1.23.0)78 in R (v.4.3). Next, we aggregated mean counts across annotated cell types with scuttle (v.1.11.2 (ref. 79); sciwheel.com/work/citation?ids=3436659&pre=&suf=&sa=0). After aggregation, we removed genes with zero expression and applied transcripts per million (TPM) normalization. Across all cell types, we calculated a specificity score for each gene defined as the proportion of total expression of a gene. To assign marker genes based on cell specificity, we filtered out genes with less than one TPM and selected the top 10% of genes based on the specificity score for each cell type. We used these marker genes to assess the enrichment of ancestry-associated DEGs using a two-sided Fisher’s exact test and corrected for multiple testing with the Benjamini–Hochberg method.
