Fig. 8From: scINSIGHT for interpreting single-cell gene expression from biologically heterogeneous dataRelative performance of scINSIGHT and the other integration methods. To evaluate the performance on clustering (A), for each dataset, the relative performance of a method was calculated as its ARI or median Silhouette score divided by the highest score on that dataset. To evaluate the performance on integration (B), for each dataset, the relative performance of a method was calculated as its integration score divided by the highest integration score on that dataset. In both cases, the method with a relative performance of 1 is the best method on a dataset. In each heatmap, the methods are ordered by the average relative performance across all datasetsBack to article page