Work

NMD_orf_model_v5_4ct: deep-learning model of nonsense-mediated decay from ORF sequence context

Year: 2026

Type: software

Abstract: Code-only, runnable repackaging of the NMD ORF model (4 cell-type retrain). Same model and results as v1.0.0 (DOI 10.5281/zenodo.21536502): Ships code only — results_4ct/ intermediates are regenerated by the pipeline (see README "Build order"); starting data are in GEO (GSE329233). Removed 10 dangling feature-input symlinks; producers retained (export_rds.R, data_prep.py, relabel_tx_summary_4ct.R). Archived superseded QC logs and redundant SLURM wrappers to superseded/. Added the uORF-attention attribution analysis and documented it in METHODS.

Source: Zenodo (CERN European Organization for Nuclear Research)

DOI: 10.5281/zenodo.21536501

Authors: Leshem, Yul, Kasai, Yoshiyuki, Eskandar, Masih, Thakur, Sanika M., Santos Cabrera, Jose, Runyan, Matthew, Paul, Ayan, Ziniti, John, Boueiz, Adel, Saferali, Aabida, De Marzio, Margherita, Dy, Jennifer, Laederach, Alain, Tata, Aleksandra, Tata, Purushothama Rao, Randell, Scott H., Castaldi, Peter J.

Institutions: Harvard University, Mass General Brigham, University of North Carolina at Chapel Hill, Northeastern University, Duke University

Language: English

Topic: RNA Research and Splicing

Subfield: Molecular Biology

Field: Biochemistry, Genetics and Molecular Biology

Domain: Life Sciences

Open Access status: green

OpenAlex ID: W7170294438

Cites:

Cited by: 1

Related to: 10