Single nuclei provide comparable clustering resolution to cells with inclusion of intronic reads. (A) Co-clustering heatmaps show the proportion of 100 clustering iterations that each pair of nuclei were assigned to the same cluster. Clustering was performed using gene expression quantified with exonic reads or intronic plus exonic reads for two key clustering steps: selecting significantly differentially expressed (DE) genes and calculating pairwise similarities between nuclei. Co-clustering heatmaps were generated for each combination of gene expression values, and blue boxes highlight 11 clusters of nuclei that consistently co-clustered using introns and exons (upper left heatmap) and were overlaid on the remaining heatmaps. The row and column order of nuclei is the same for all heatmaps. (B) Co-clustering heatmaps were generated for cells as described for nuclei in (A), and blue boxes highlight 11 clusters of cells. (C) Cluster cohesion (average within cluster co-clustering) and separation (difference between within cluster co-clustering and maximum between cluster co-clustering) are plotted for nuclei and cells and all combinations of reads. Including introns in gene expression quantification dramatically increases cohesion and separation of nuclei but not cell clusters.