The transitional cerebellar progenitor is a technical artifact driven by low-quality cells [re-analysis]

We reprocessed the raw FASTQ files from the published single-cell RNA-seq atlas of the developing human cerebellum (accession GSE198565) and, following rigorous quality control, removed low-quality cells and refined the cell-type annotations. Both raw scRNA-seq FASTQs and cellranger mapping outputs from the fetal human cerebellar atlas (accession GSE198565, sample GSM5952337), were obtained and reprocessed using the same parameters described in the publication. FASTQ files for GSM5952337 were provided by the submitter. FASTQ files were mapped with CellRanger (V3.2, the same as the cellranger outputs provided) and cell type annotations were adopted from the original publication (GSE198565). The dataset was composed of 95,542 cells collected from fetal cerebella of eight individual donors. Additional QC metrics were calculated for each sample utilizing CELLBENDE and DropletQC. UMAP coordinates were all calculated using cosine distance between the top 100 nearest neighbors based on the first 50 principal components. Unsupervised clustered were determined using leiden clustering at a resolution of 0.8 as implemented in scanpy.