Fig. 3
From: Analytic Pearson residuals for normalization of single-cell RNA-seq UMI data
t-SNE embeddings of three retinal datasets. Panels in each column are based on a different data transformation method with PCA or GLM-PCA reduction to 50 dimensions (see “Methods”), and each row shows a different retinal dataset. We did not perform any gene selection here. Colors correspond to cell type labels provided by the original papers. a–c Full-retina dataset (DropSeq) [33], containing all retinal cell types (including glia and vascular cells). 24, 769 cells. d–f Bipolar cell dataset (DropSeq) [37]. 13,987 cells. g–i Retinal ganglion cell dataset (10X v2) [38]. 15,750 cells