Aging causes changes in transcriptional noise across a diverse set of cell types

Aging and its associated diseases result from complex changes in cell state which can be examined with single-cell transcriptomic approaches. We analyzed gene expression noise, a measure of cellular heterogeneity, across age and many cell types and tissues using the single cell atlas Tabula Muris Senis , and characterized the noise properties of most coding genes. We developed a quantitative, well-calibrated statistical model of single-cell RNAseq measurement from which we sensitively detected changes in gene expression noise. We found thousands of genes with significantly changing gene expression noise with age. Not all genes had increasing noise with age—many showed a robust decreases of noise. There were clear biological correlation between subsets of genes, with a systemic decrease of noise in oxidative phosphorylation pathways while immune pathways involved in antigen presentation saw an increase. These effects were seen robustly across cell types and tissues, impacting many organs of healthy, aging mice.