Universal preprocessing of single-cell genomics data

Large single-cell atlases depend on comparing datasets generated by many assays, so preprocessing must be consistent before biological interpretation begins. Existing workflows are often assay-specific, while read structures, barcode correction, and UMI/read counting differ across RNA, ATAC, protein, tags, and spatial modalities. cellatlas uses machine-readable seqspec files to configure modular tools for uniform preprocessing across nine datasets, modalities, and single-cell technologies.