Pervasive transcription = genes + noise

Most of the DNA in the human genome is transcribed at some point in development or in some cell type. This fact has been known since the late 1960s.

There are basically two types of transcripts. Functional transcripts mostly come from genes although there might be a few exceptions (e.g. enhancer RNAs). Non-functional transcripts can be produced by pseudogenes or from virus and transposon fossils. They can also due to transcriptional noise caused by spurious transcription.

It's surprisingly difficult to calculate the amount of the genome occupied by genes. One problem is that there are many different estimates of the number of non-coding genes. (The number of protein-coding genes is known to be close to 20,000.) Another problem is that the sizes of introns and exons aren't known with any precision and that's partly due to problems with annotation.

Nevertheless, we can make a reasonable estimate and that's what I did in my book in chapter 6: "How Many Genes? How Many Proteins?" I took the best estimates of coding DNA, UTRs, and introns and calculated that the average size of a protein-coding gene was 61,760 bp (61.8 Kb). Assuming that there are 19,500 of these genes, they should occupy 39% of the genome. Adding in some reasonable (and generous) estimate of non-coding genes brings the total to about 45% of the genome.¹

Thus, 45% of the genome should be transcribed to produce functional RNAs. We don't know exactly how many pseudogenes are still transcribed and we don't know how many defective transposons and virus DNA are still transcribed but it's safe to say that it might represent 5% of the genome. That means we can account for 50% of the genome being transcribed to produce easily detectable RNAs. That alone, accounts for pervasive transcription.

However, many studies report that at least 75% of the genome is transcribed and the highest estimates are over 80%. The other transcripts are almost always present at very low concentrations and are rapidly degraded. The DNA sequences that are transcribed are not conserved. This strongly suggests that these are spurious transcripts. We know that such transcripts exist and we know that random sequences of DNA will be transcribed by accident giving rise to junk RNA.

A newly published paper has attempted to sort out the amount of DNA that produces functional RNAs and the amount that produces junk RNA (noise).

Adey, B.N., Maddock, D.J., Hermann-Le Denmat, S., Dinger, M.E., Gardner, P.P., Poole, A.M. and Ganley, A.R.D. (2026) Pervasive transcription in the human genome exceeds background noise. Genome Biology and Evolution:evag042. [doi: 10.1093/gbe/evag042]

Large genomes such as the human genome are pervasively transcribed yet encode relatively few unambiguously functional elements. This has led to debate over whether pervasive transcription is indicative of large suites of uncharacterized functional elements or is simply background noise. Here we used a deep-learning model to estimate background transcription in the human genome as a way of distinguishing between these two hypotheses. We applied the model to randomised (reversed or shuffled) versions of the human genome and found that transcription is predicted to be sparse across all randomisation methods, initiating with at least four-fold lower frequencies than in the native human genome. This relatively low level of background transcription from the human genome suggests that most transcription is not a consequence of background noise, thus it requires other explanations. We find that randomizing only interspersed repeats in human genome has little impact on predicted transcription, suggesting that transcription of mobile elements does not explain the excess transcription in the human genome. Instead, most transcriptional events may derive from functional noncoding RNA transcripts, some general requirement for extensive transcription initiation/elongation, and/or mutational biases leading to the frequent appearance of transcription initiation sites by chance.

The authors assume that 75% of the genome is transcribed and they are interested in knowing how much of this is due to large numbers of potential non-coding genes. They clam that their results are "an important step in resolving the long-standing debate over what proportion of the human genome is functional." They developed models to estimate the extent of background noise (spurious transcription) from random DNA sequences and their model leads them to conclude that spurious transcription cannot account for most of that 75%.

This isn't news. As I've shown above, transcription of known genes accounts for about 45% of the genome or most of pervasive transcription. What's interesting is that the authors don't discuss introns and don't seem to recognize that a lot of pervasive transcription is from introns. In fact, the authors conclude that gene transcription cannot account for a large fraction of pervasive transcription.

In summary, we found that transcription initiates in in silico randomized human genomes at only about one-quarter of the level observed in the native human genome. This disparity is unlikely to be explained solely by the presence of genes in the native genome, as the number of TSS clusters in the T2T forward genome (742,037 clusters with a maximum cluster gap of 25 bp) is vastly greater than the number of genes. On the basis of these results, we conclude that the level of background transcription initiation activity in the human genome is relatively low and does not account for the majority of the transcription in the human genome.

What's going on here? We already knew that spurious transcription (background transcription) accounts for a lot of transcribed DNA but genes account for most of the transcribed genome. In my opinion, there's nothing new in this paper that's worth publishing. Am I completely off base in my calculations or has the peer review system failed us again?

---

1. Keep in mind that there are hundreds of ribosomal RNA genes and they are quite large.