SpliceSelectNet: a hierarchical Transformer-based deep learning model for splice site prediction

Yuna Miyachi, Kenta Nakai · 2026

Read SpliceSelectNet: a hierarchical Transformer-based deep learning model for splice site prediction by Yuna Miyachi, Kenta Nakai (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/pmc-13284721.txt.

Description

Yuna Miyachi and Kenta Nakai built a deep learning model in 2026 that predicts where RNA gets spliced, reading DNA stretches up to one hundred thousand letters long. Splicing trims non-coding segments from gene messages, and errors cause disease. Older tools miss far-away control signals because long sequences cost too much to process, and many models cannot explain their calls.

Their model, SpliceSelectNet, stacks transformer layers in a hierarchy, mixing local and global attention to catch near and distant signals at single-letter precision. It topped benchmark tests for splice site prediction and for spotting harmful splicing changes. Systematic fake mutations showed the model's attention scores track real functional importance, and long-range perturbation tests confirmed it senses distant regulators. The authors present it as both accurate and biologically readable.

Catalog record

CREATORYuna Miyachi, Kenta Nakai
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE91 KB
VERSION1.0
LICENSECC BY 4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/articles/PMC13284721

Browse the full library · About the Tower of Babel