Math & SciencesOperationsOpen accessPublished 3 Oct 2026
Run the Parithera scanpy single-cell pipeline — load and concatenate samples, QC (mt/ribo/hb, scverse MAD-based outlier filtering + mito cutoff, Scrublet doublets), normalize + log1p, highly variable genes, PCA, UMAP and t-SNE, Leiden clustering at several resolutions, and Wilcoxon differential expression — on one or more samples, then render the result as an interactive two-panel UMAP (Leiden clusters vs sample) with box/lasso selection, gene/UMI views, and saved, reusable cell selections. Use when the user wants to process single-cell samples end to end, cluster them, explore or gate cells, or compare clusters across samples. Requires raw per-sample matrices (.h5ad / 10…