EUVIMEDEUVIMED entdecken
Open Access · CC BY-NC-ND 4.0

ADAR-GPT : A continually fine-tuned language model for predicting A-to-I RNA editing sites

Zohar Rosenwasser, Roni Cohen-Fultheim, Michael Levitt iD, Erez Y Levanon, Gal Oren

Proceedings of the National Academy of Sciences · 13.01.2026 · 123 (2)

DOI: 10.1073/pnas.2529073123 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the National Academy of Sciences, online erschienen am 09.01.2026. Verfasst von Zohar Rosenwasser und 4 weiteren Autor:innen. Beteiligte Einrichtung: Faculty of Life Sciences, The Mina and Everard Goodman, Ilan University, Ramat Gan 5290002, Israel. Hauptthemen laut MEDLINE: RNA Editing, Inosine, Adenosine Deaminase, Adenosine. Die Arbeit wurde bisher 3-mal zitiert (Crossref) und nennt 77 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0). Gefördert von: NIGMS NIH HHS, HHS | NIH | National Institute of General Medical Sciences (NIGMS), Israeli Ministry of Science.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Adenosine-to-inosine (A-to-I) RNA editing by ADAR enzymes shapes transcript fate and underpins emerging RNA editing therapeutics, yet predicting which adenosines are edited remains difficult. We introduce ADAR-GPT, a model-agnostic fine-tuning framework that adapts a GPT-class language model to classify editing at candidate sites using sequence context in standardized 201 nt windows with the target adenosine explicitly marked. We train and evaluate on GTEx liver data ( n = 131 samples) at a clinically relevant 15% editing threshold, using a two-stage continual fine-tuning approach where lower thresholds serve as curriculum data to progressively sharpen decision boundaries. Using sequence data, ADAR-GPT demonstrates competitive or superior performance when benchmarked against established computational approaches, including convolutional and foundation model architectures, achieving a better balance of recall, precision, and specificity alongside stronger operating-curve metrics. The approach is reproducible and portable across GPT backbones without architectural changes. Beyond accurate site classification, ADAR-GPT provides practical adenosine scoring to prioritize experimental targets and inform guide RNA design, with a framework adaptable to new datasets and model architectures.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

RNA EditingInosineAdenosine DeaminaseAdenosineLarge Language ModelsHumansRNA-Binding ProteinsComputational Biology

Stichwörter der Autor:innen

ADAR enzymesRNA editingRNA structurecontinual fine-tuningfoundation models

Substanzen

InosineAdenosine DeaminaseAdenosineRNA-Binding Proteins

Autor:innen und Institutionen

  • Faculty of Life Sciences, The Mina and Everard Goodman, Ilan University, Ramat Gan 5290002, Israel.
  • Department of Structural Biology, School of Medicine, Stanford University, Stanford, CA 94305.
  • Computer Science Department, Technion-Israel Institute of Technology, Haifa 32000, Israel.

ORCID-Profile: Michael Levitt

Förderung

  • NIGMS NIH HHS
  • HHS | NIH | National Institute of General Medical Sciences (NIGMS)
  • Israeli Ministry of Science
  • Israel Science Foundation (ISF)
  • Foundation Fighting Blindness

Zitieren

APA

Rosenwasser, Z., Cohen-Fultheim, R., Levitt, M., Levanon, E. Y., & Oren, G. (2026). ADAR-GPT : A continually fine-tuned language model for predicting A-to-I RNA editing sites. Proceedings of the National Academy of Sciences, 123(2). https://doi.org/10.1073/pnas.2529073123

Vancouver

Rosenwasser Z, Cohen-Fultheim R, Levitt M, Levanon EY, Oren G. ADAR-GPT : A continually fine-tuned language model for predicting A-to-I RNA editing sites. Proceedings of the National Academy of Sciences. 2026;123(2). doi:10.1073/pnas.2529073123

BibTeX
@article{Rosenwasser2026,
  author = {Rosenwasser, Zohar and Cohen-Fultheim, Roni and Levitt, Michael and Levanon, Erez Y and Oren, Gal},
  title = {ADAR-GPT : A continually fine-tuned language model for predicting A-to-I RNA editing sites},
  journal = {Proceedings of the National Academy of Sciences},
  year = {2026},
  volume = {123},
  number = {2},
  doi = {10.1073/pnas.2529073123}
}
Kostenlos starten →