Comparison of nuclear and whole cell transcriptomes. (A) Left: Percentage of RNA-seq reads mapping to genomic regions for cells, nuclei, and whole brain control RNA. Bars indicate median and 25th and 75th quantiles. Note that among cells exonic and intronic read alignment is bimodal. Right: Gene detection (counts per million, CPM > 0) based on reads mapping to exons, introns, or both introns and exons. (B) Left: The most similar pair of cells have more highly correlated gene expression (r = 0.92) than the most similar pair of nuclei (r = 0.76), due to fewer gene dropouts. Right: Cells have consistently more similar expression to each other than nuclei, even after correcting for gene dropouts based on an expression noise model. (C) Left: Binned scatter plot showing all genes are detected (CPM > 0) with equal or greater reliability in cells than nuclei. Grey lines show the variation in detection that is expected by chance (95% confidence interval). Right: Binned scatter plot showing 0.4% of genes are significantly more highly expressed (fold change > 1.5, adjusted P-value < 0.05) in nuclei , and 20.5% of genes are more highly expressed in cells. The log-transformed color scale indicates the number of genes in each bin. (D) Nuclear enriched genes are highly enriched for genes involved in neuronal connectivity, synaptic transmission, and intrinsic firing properties. Cell enriched genes are predominantly related to mRNA processing and protein translation and degradation. In addition, immediate early gene expression is increased up to 10-fold in cells, despite comparable isolation protocols for cells and nuclei.