Fig. 6
From: SCA: recovering single-cell heterogeneity through information-based dimensionality reduction
a UMAP plots downstream of PCA and SCA on CD34+ immune cell data with 7 cell types and 3 donors. With further iterations, batches appear more integrated (top) whereas cell types remain separate (bottom). b Silhouette scores of the donor and cell type groupings, using Euclidean distance downstream of PCA or SCA. The donor silhouette score decreases with further iterations of SCA (left), whereas the cell type silhouette score remains above that of PCA (right)