Fig. 4

Differential expression and abundance analysis using supercells. A, B Heatmaps illustrate the scaled and centred median expression of cell state markers, calculated for each sample across the supercells, for B cells and CD4 T cells in the BCR_XL dataset. Each sample (row) is annotated according to its stimulation status, either with B cell receptor / FC receptor cross-linker (BCR-XL) or unstimulated. Each marker (column) is annotated based on its statistical significance as determined by Limma at a 5% False Discovery Rate (FDR_sig). C-F Differential abundance analysis following supercell creation for the Anti_PD1 dataset. C A Multidimensional Scaling (MDS) plot showcasing the variance in samples following cyCombine correction. D Distribution of marker expression for cyCombine-corrected supercells. E The proportion of single cells for the rare monocyte subset cluster (cluster_10). F Comparison of FDR obtained by running Propeller at the single cell or supercell level. The y-axis represents the -log10 transformed FDR, with lower FDR (more significant) corresponding to higher -log10 values. The red dotted line shows an equivalent FDR of 0.1