Fig. 3: Comparison of AUROC for identifying cell type markers across 9 methods using 10 real datasets (Tabula Muris and Azimuth demo datasets). | Communications Biology

Fig. 3: Comparison of AUROC for identifying cell type markers across 9 methods using 10 real datasets (Tabula Muris and Azimuth demo datasets).

From: Cluster-independent multiscale marker identification in single-cell RNA-seq data using localized marker detector (LMD)

Fig. 3

Each block represents the AUROC score for a method on a dataset, with the number showing the AUROC value and the color indicating the method’s rank. Marcopolo (CPU) was only run on the Tabula Muris dataset due to runtime and hardware restrictions. Two true marker sets: 1 genes listed in marker databases (CellMarkerDB and Azimuth marker list); 2 the top 100 genes with the maximum log fold change across pre-annotated cell types.

Back to article page