Fig. 5: SimiC’s proposed scores capture cell-state specific regulatory dynamics on the CD8+ CAR T cell dataset52.
From: SimiC enables the inference of complex gene regulatory dynamics across cell phenotypes
a Heatmap depicting the regulatory dissimilarity between the two cell phenotypes (IP and D12) for different regulons for CD8+ exhausted, CD8+ effector, and CD8+ memory cells. Violin plots of the activity scores for the different cell states, as well as the tSNE plots showing the CAR T cells of different cell states colored by their activity score for the MYC regulon as calculated by SimiC (b) and SCENIC (c); for the RUNX3 regulon as calculated by SimiC (d) and ICANet (e); and for the EOMES regulon as calculated by SimiC (f) and SCENIC (g). h Cell state clustering performance as measured by the ARI score. For SCENIC we consider the binarized and non-binarized scores. A larger ARI indicates a better clustering performance. SINCERITIES was left out of the analysis as it needs at least four phenotypes to run.
