The abstract is read at the publisher; the summary is JClub's.
A deep learning model trained on ~2 million synthetic 3'UTR inserts in A549 and HCT116 cells captures most RNA output and nuclear–cytoplasmic partitioning behavior, with context dependence limited to rare extremes.
- Why it matters: Sequence rules for how 3'UTRs control transcript abundance and localization are still incompletely defined, limiting interpretation of natural variants and disease links.
- What they did: The study measured RNA output and nuclear–cytoplasmic partitioning from ~2 million synthetic 3'UTR inserts in A549 and HCT116, identifying short 2- to 8-nt elements and fitting a context-aware deep learning model.
- The result: Applied to ClinVar 3'UTR variants, the model prioritizes a subset of pathogenic SNVs with aberrant predicted effects, often explained by creation or disruption of splice-site-like U1 motifs linked to telescripting-associated readthrough.
The findingWhy it mattersWhat they didThe result
AI summary of the abstract · may contain errors · not medical advice · read the paper.