Extended Data Fig. 1: Evaluating sample quality.

(A-C) Quality control workflow and resulting UMAPs with cells colored by manually assigned cell type IDs based on the expression of hallmark skeletal muscle genes. Prior to any quality-control there were 365,710 cells and 24 cell types were identified, including 6 low-quality (LQ) clusters (A). After ambient RNA removal with SoupX, 23 cell types were identified, including 5 LQ clusters (B). After ambient RNA removal with SoupX and removal of LQ cells based on the number of genes and UMIs and the percent of mitochondrial reads, there were 286,273 cells and 24 cell types were identified, including two doublet clusters (C). (D) This is the same UMAP as in (B), but the cells are colored by quality status. Cells that had <200 genes, <750 UMIs, and >25% mitochondrial reads are considered LQ. All other cells are considered high-quality (HQ). (E) This is the same UMAP as in (C), but the cells are colored by doublet status as determined by DoubletFinder using an estimated doublet rate of 5%. (F) For every cell type cluster in (C), the fraction of singlets and doublets was calculated. (G-I) Violin plots of the number of genes (G), the number of UMIs (H), and the percent of mitochondrial reads (I) in each sample.