Deep learning to decode sites of RNA translation in normal and cancerous tissues

Sep 21, 2026

RNA translation is essential for cellular life and strongly influences human disease, yet reliably defining variation in translation remains a major analytical challenge. The work introduces RiboTIE, a transformer model that maps global RNA translation with unmatched precision and sensitivity for ribosome profiling data. Applying RiboTIE to normal brain and medulloblastoma samples provides high-resolution views of disease regulation, revealing how translation control differs in cancerous tissue.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages22

Reading Time55 min

AccessFree to download

Abstract

RiboTIE transformer-based mapping approach

Global insights from ribosome profiling (Ribo-Seq)

Challenges and biases in Ribo-Seq analyses

Model features and interpretation

bioRxiv preprint doi: [https://doi.org/10.1101/2024.03.21.586110](https://doi.org/10.1101/2024.03.21.586110); this version posted March 25, 2024. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license. Deep learning to decode sites of RNA transla􀆟on in normal and cancerous 􀆟ssues Jim Clauwaert1-3,8, Zahra McVey4, Ramneek Gupta4, Ian Yannuzzi5, Gerben Menschaert6,7,8 & John R. Prensner1-3,7,8 Aﬃlia􀆟ons 1 Division of Pediatric Hematology/Oncology, Department of Pediatrics, University of Michigan, Ann Arbor, MI, USA 2 Chad Carr Pediatric Brain Tumor Center, University of Michigan, Ann Arbor, MI, USA 3 Department of Biological Chemistry, University of Michigan, Ann Arbor, MI, USA 4 Novo Nordisk Research Centre Oxford, Novo Nordisk Ltd., Oxford, United Kingdom 5 Cancer Program, Broad Ins􀆟tute of MIT and Harvard, Cambridge, MA, USA 6 Department of Data Analysis and Mathema􀆟cal Modelling, Ghent University, Belgium 7 These authors share senior authorship: Gerben Menschaert, John R. Prensner 8 These authors are corresponding authors: Jim Clauwaert, Gerben Menschaert, John R. Prensner Key words: Ribo-seq, non-canonical open reading frames, cancer, medulloblastoma, RNA transla􀆟on Address correspondence to: John R. Prensner, MD, PhD Department of Pediatrics and Biological Chemistry, University of Michigan Medical Science Research Building II, Room 2560B 1150 Medical Center Drive Ann Arbor, MI, 48109 Email: [prensner@umich.edu](prensner@umich.edu) Phone: 734-763-5939 Jim Clauwaert, PhD Department of Pediatrics and Biological Chemistry, University of Michigan Medical Science Research Building II, Room 2560B 1150 Medical Center Drive Ann Arbor, MI, 48109 Email: [clauwaer@umich.edu](clauwaer@umich.edu) Phone: 734-763-9031 Gerben Menschaert, PhD Department of Data Analysis and Mathema􀆟cal Modelling, Ghent University Coupure Links 653 900 Gent, Belgium Email: gerben.menschaert@ugent.be Phone: +32-926-49922 bioRxiv preprint doi: [https://doi.org/10.1101/2024.03.21.586110](https://doi.org/10.1101/2024.03.21.586110); this version posted March 25, 2024. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license. Abstract The biological process of RNA transla􀆟on is fundamental to cellular life and has wide-ranging implica􀆟ons for human disease. Yet, accurately delinea􀆟ng the varia􀆟on in RNA transla􀆟on represents a signiﬁcant challenge. Here, we develop RiboTIE, a transformer model-based approach to map global RNA transla􀆟on. We ﬁnd that RiboTIE oﬀers unparalleled precision and sensi􀆟vity for ribosome proﬁling data. Applica􀆟on of RiboTIE to normal brain and medulloblastoma cancer samples enables high-resolu􀆟on insights into disease regula􀆟on of RNA transla􀆟on. Word count: 75 bioRxiv preprint doi: [https://doi.org/10.1101/2024.03.21.586110](https://doi.org/10.1101/2024.03.21.586110); this version posted March 25, 2024. The copyright holder for this preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under aCC-BY-ND 4.0 International license. Main RNA transla􀆟on is an intricate process that involves the stepwise binding of the 40S and 60S ribosome subunits to RNA, along with mul􀆟ple eukaryo􀆟c ini􀆟a􀆟on factors and other cofactors.1 RNA transla􀆟on is a major determinant of protein abundance and represents a core area of disease biology including cancer, where numerous gene􀆟c and non-gene􀆟c factors alter the composi􀆟on of ribosomes, eﬃciency of transla􀆟on, and ﬁdelity of transla􀆟on.2 To gain global i

FAQs

What is the main goal of the study?

To accurately map global RNA translation by decoding translation sites using a transformer model approach.

How does RiboTIE differ from prior Ribo-Seq analysis tools?

It uses transformer-based learning to extract dataset-dependent correlations and scores candidate translation initiation sites across transcripts without requiring a pre-processing step.

What data and inputs does RiboTIE rely on?

It uses sequenced and mapped ribosome-protected fragments (RPFs) along transcripts and does not use DNA sequence information such as start codons or ORF characteristics.

Sep 21, 2026

Uploaded by Olivia Brown