Persistent Features of Intermittent Transcription

Abstract

Here we report statistical studies of single-cell mRNA counts from cells derived from different tissues of adult mice. By examining correlations between mRNA gene counts we find strong evidence that when genes are only observed in a small fraction of cells, this is as a consequence of intermittent transcription rather than of expression only in specialized cell types. Count statistics are used to estimate a peak transcription level for each gene, and a probability for the gene to be active in any given cell. We find that the peak transcription levels are approximately constant across different tissue types, but the gene expression probabilities may be markedly different. Both these quantities have very wide ranges of values, with a probability density function well approximated by a power law.\n\nAuthor summaryUsing evidence from single-cell mRNA counts, we argue that the expression of many genes in individual mouse cells is highly intermittent. Comparing cells from different tissues, we find that the peak activity of a given gene is approximately the same in all tissue types, whereas the probability of a gene being active can differ markedly.