Abstract

Transposed elements (TEs) typically comprise 30-50% of the mammalian genome and, until recently, their repetitive composition has restricted our ability to study individual TEs. Here I will present our early statistical modelling that indicated high throughput sequencing could be successfully applied to detect transcription from specific TEs, as well as a computational solution to the related multi-map tag phenomenon. These works formed the basis for a subsequent landmark publication proving that TEs were heavily transcribed in mammals. The associated transcripts, dubbed teRNAs, supplied up to 30% of the capped RNA molecules in cancer, germ and somatic cells. An analysis of 250 000 TE promoter regions revealed that these frequently functioned as alternative promoters for nearby genes and/or expressed ncRNAs. Overall, we discovered that the abundance of teRNAs far exceeded prior estimates, with a common theme of strong tissue specificity and association with nearby protein-coding genes, demonstrating that TEs were an essential part of the mammalian transcriptome and regulome.