Extended Data Fig. 2: The comparison between pseudo-cell and single-cell of MOCA as the reference of label transfer.
From: A single-cell atlas of chromatin accessibility in mouse organogenesis
a, UMAP visualization of MOCA pseudo-cell data of 43 main cell types showing improved cell-type clustering (n = 400 pseudo-cells in each cell type), colored by cell-type assignment. Random sampling 5 cells in the same cell type from the MOCA dataset were aggregated to make one pseudo-cell (See Method). b, Uniform manifold approximation and projection (UMAP) visualization of integrated results of subsampling cells from scATAC-seq (this study) and pseudocells from scRNA-seq (TOME) for the 43 major cell types (see Methods), colored by omics (right) and cell-type assignment (left). c, Violin plot showing the number of genes and unique molecular identifiers (UMIs) in raw single cells and pseudo-cells. d, Confusion matrix comparing the annotation of scATAC cells using marker genes in Fig. 1c and transferred labels of scATAC cells with the nearest scRNA-seq cell following the integration of the scATAC and raw MOCA single-cell dataset (up) or and pseudocell scRNA-seq dataset (bottom). Coloring indicates the proportion of cells mapping for a given pair. e, Density plot showing the distribution of predicted scores of each cell calculated by the label transfer algorithm in Seurat packages. Red and cyan dotted lines indicate the median predicted scores of pseudocell and single cell as a reference, respectively. Source numerical data are provided as source data.
