Back

A systematic delineation of 3'UTR regulatory elements and their contextual associations

Liang, N.; Li, J.-Y.; Ding, Y.; Wang, Y.; Champer, J.; Ren, L.-C.; Gao, G.

2025-06-09 genomics
10.1101/2025.06.09.658412 bioRxiv
Show abstract

The 3'UTR encodes regulatory information that shapes transcript abundance and subcellular distribution, yet the underlying sequence rules remain incompletely defined. Using SEERS, we quantified the effects of [~]2 million synthetic 3'UTR inserts in A549 and HCT116 cells on RNA output and nuclear-cytoplasmic partitioning. We identify a broad repertoire of short (2[~]8 nt) elements whose effects largely align along a major coupled axis that links high expression with cytoplasmic enrichment and low expression with nuclear enrichment, and contribute predominantly in an additive manner. A context-aware deep learning model (TALE) captures most of this behavior while revealing that strong context dependence is uncommon and emerges mainly in rare, extreme contexts consistent with higher-order constraints. Applying TALE to ClinVar 3'UTR variants prioritizes a subset of pathogenic SNVs with aberrant predicted effects, frequently explained by creation or disruption of splice-site-like U1 telescripting signals, highlighting a potent route to 3'UTR-driven disease mechanisms.

Matching journals

The top 3 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.