Every study brought into scEiaD was hand curated to identify the unique biological samples, published paper ID (where available), organ, tissue (e.g., cornea), source (tissue), single-cell platform (e.g., 10X v2, DropSeq). We also, where possible, annotated retina region (e.g., macula), sex, and age or developmental stage. Samples with perturbations (genetic or treatment) were not included. To import individual cell-type labels (e.g., rod, Müller glia), we wrote custom code for each study that made this table available to link their cell-type assignments back to the matching barcoded cell. We curated two types of cell type labels: CellType and SubCellType, where the former is our normalized cell types taken from the published labels and the latter are the original cell-type labels. We normalized names for CellType—for example, changing “MG” to “Muller Glia” and dropping more detailed cell type assignments, such as off and on bipolar cells for Bipolar Cells (as few studies went into this level).