Due to the sparsity of single-cell data, differential expression methods designed to be run on the single-cell level often lack high statistical power. To account for this challenge, we utilized a pseudobulk approach. To create the pseudobulk data, for each cell type, the gene expression matrices of each cell of that cell type were combined (summed) by sample ID. Samples were removed on a cell type-specific basis if the sample contained less than 50 cells of that cell type. See Supplementary Table 7 for a summary of the number of pseudobulk samples created for each cell type. Due to a low number of samples (less than 10 individuals with AUD and 10 without) meeting the >=50 cell criteria, differential gene expression analysis was not performed for cholinergic interneurons, vascular smooth muscle cells, CCK interneurons, and macrophages. These pseudobulk samples were used for the differential expression analysis, below.
