A single-cell transcriptomic dataset of pluripotent stem cell derived STGD1 disease model

Quality control and filteringThe matrices were converted to Seurat objects individually using the Seurat R package (version 5.1.0). The data were filtered to remove cells with fewer than 200 unique genes per cell or greater than 10000 genes per cell, or more than 15% reads mapped to mitochondrial genes. Filtered gene matrices were normalized with the ‘NormalizedData’ function using the ‘LogNormalize’ method. The top 2000 variable genes were identified using the ‘vst’ method in the ‘FindVariableFeature’ function.Integration, dimensional-reduction and clusteringBatch effects between samples were corrected using the standard integration algorithm with “FindIntegrationAnchors” function in Seurat to minimize the influence of technical variability. Dimensionality was reduced using principal component analysis (PCA). In this way, a neighbourhood graph was calculated using 15 principal components. Following a Uniform Manifold Approximation and Projection (UMAP) embedding, cell clusters were detected based on the Louvain clustering implementation using a resolution parameter of 0.25.Cell-type annotationTo facilitate the cell type annotation, DGEs was calculated by using ‘FindAllMarkers’ function on normalized count data to identify markers of each cluster. Cell identities were assigned based on canonical markers assembled from relevant literature.