PT  - JOURNAL ARTICLE
AU  - Mathys Grapotte
AU  - Manu Saraswat
AU  - Chloé Bessière
AU  - Christophe Menichelli
AU  - Jordan A. Ramilowski
AU  - Jessica Severin
AU  - Yoshihide Hayashizaki
AU  - Masayoshi Itoh
AU  - Michihira Tagami
AU  - Mitsuyoshi Murata
AU  - Miki Kojima-Ishiyama
AU  - Shohei Noma
AU  - Shuhei Noguchi
AU  - Takeya Kasukawa
AU  - Akira Hasegawa
AU  - Harukazu Suzuki
AU  - Hiromi Nishiyori-Sueki
AU  - Martin C. Frith
AU  - FANTOM consortium
AU  - Clément Chatelain
AU  - Piero Carninci
AU  - Michiel J.L. de Hoon
AU  - Wyeth W. Wasserman
AU  - Laurent Bréhélin
AU  - Charles-Henri Lecellier
TI  - Discovery of widespread transcription initiation at microsatellites predictable by sequence-based deep neural network
AID  - 10.1101/2020.07.10.195636
DP  - 2020 Jan 01
TA  - bioRxiv
PG  - 2020.07.10.195636
4099  - http://biorxiv.org/content/early/2020/07/10/2020.07.10.195636.short
4100  - http://biorxiv.org/content/early/2020/07/10/2020.07.10.195636.full
AB  - Using the Cap Analysis of Gene Expression (CAGE) technology, the FANTOM5 consortium provided one of the most comprehensive maps of Transcription Start Sites (TSSs) in several species. Strikingly, ~ 72% of them could not be assigned to a specific gene and initiate at unconventional regions, outside promoters or enhancers. Here, we probed these unassigned TSSs and showed that, in all species studied, a significant fraction of CAGE peaks initiate at microsatellites, also called short tandem repeats (STRs). To confirm this transcription, we developed Cap Trap RNA-seq, a technology which combines cap trapping and long reads MinION sequencing. We trained sequence-based deep learning models able to predict CAGE signal at STRs with high accuracy. These models unveiled the importance of STR surrounding sequences not only to distinguish STR classes, as defined by the repeated DNA motif, one from each other, but also to predict their transcription. Excitingly, our models predicted that genetic variants linked to human diseases affect STR-associated transcription and correspond precisely to the key positions identified by our models to predict transcription. Together, our results extend the repertoire of non-coding transcription associated with DNA tandem repeats and complexify STR polymorphism.Competing Interest StatementThe authors have declared no competing interest.