Figure 3.
TSDstrat removes CC-dependent confounder genes, which are not recognized by current transcriptome-based approaches. (A) CC progression can be transcriptionally inferred and corrected with regression by the R library Seurat. These inferred CC phases correlate with TSD. (B) TSDstrat removes effects of CC heterogeneity from candidate identification. Heatmap of top 50 NOstrat, TSDstrat, KINstrat, and trackSeq candidates’ expression per cell. Only NOstrat columns and rows cluster for CC progression. Thus, TSD sufficiently controls for CC heterogeneity. (C) Principal component (PC) analysis of scRNA-Seq profiles before and after CC regression. (D) TSDstrat removes different and more CC-dependent genes than CC regression.