Fig. 3: Application of PACS to the mouse kidney dataset.
From: Depth-corrected multi-factor dissection of chromatin accessibility for scATAC-seq data with PACS
a–d UMAP dimension reduction plots of the kidney adult dataset. Panels (a, c) are constructed using all features, whereas panels (b) and (d) are constructed after excluding features significantly affected by batch effects. The (a, b) are colored by batch labels, while (c, d) are colored by cell types. Features impacted by batch effects are identified using PACS (two-sided test) with FDR correction for multiple testing. Normalized PCA mixing represents the normalized mixing score calculated in the PCA space, with 1 indicating no batch effect and 0 indicating the strongest batch effect. e IGV plots of peak summits near genes specific to PCT (proximal convoluted tubule) and PST (proximal straight tubule) cell types identified using PACS. Gene specificity is determined through GREAT enrichment analysis of differentially accessible peaks. f Heatmap of normalized gene expression z-scores for the scRNA-seq data from the male (-m) and female (-f) mouse kidneys. The gene list corresponds to those identified in (e), emphasizing the consistency across datasets in identifying sex-specific expression patterns.
