Fig. 9
From: C-ziptf: stable tensor factorization for zero-inflated multi-dimensional genomics data
Condition specific GEPs recovered by C-ZIPTF in the SLE dataset: a UMAP of immune cells in SLE dataset after downsampling, colored by condition, b sample mode loading for factors 11 and 21 grouped by disease status, c 3 condition specific factors highlighted for rank 22 (Each row represents a factor, and the first three columns display the three modes: sample, cell type, and gene. The y-axis in the sample and cell type modes represent the loading of the sample or cell type on that factor. The gene mode exhibits the top 20 genes associated with the factor.), d heatmap of normalized gene expression of the top genes associated with the 3 condition specific factors in a randomly sampled set of single cells sorted by cell type and disease status