An RNA Language Model trained on sequence alone reveals the structural logic of Internal Ribosome Entry Sites
- 1Harvard Medical School
Abstract
Millions of RNA sequences are readily available, but the structures that determine their function are not. Picornaviruses initiate translation through Internal Ribosome Entry Sites (IRESes), RNA elements that recruit ribosomes independent of the 5` cap. These elements are large and highly divergent, and structural understanding remains limited to a handful of cases. We address this bottleneck by introducing an RNA language model (Albatross), trained purely on sequence, that predicts high-quality IRES structures at scale. We collect in cellulo chemical probing data for 96 full-length IRESes from divergent viruses and show that Albatross achieves far higher precision (0.80) than state-of-the-art predictions (0.47). Analyzing 75,000 IRES structures, we discover a novel Type II structural subclass and validate it experimentally. We demonstrate the pipeline broadly generalizes to identify functional structures, including tertiary contacts and alternative riboswitch structures. These findings establish Albatross as a scalable framework that accelerates RNA structure discovery and antiviral targeting.
Related
- A B1 SINE element within the mouse Malat1 long noncoding RNA acts as an internal ribosome entry site mediating micropeptide translation
- Translation initiation by the Kozak mRNA sequence is based on a conformational readout on the ribosome
- AmyloDeep: protein language model-based ensemble model for predicting amyloid propensity from the amino acid sequence
- AIPFusion-Net: an interaction-fusion framework for anti-inflammatory peptide prediction using protein language model embeddings, local sequence patterns, and handcrafted descriptors.
- Efficient structural information integration into protein language models with implicit mixture-of-experts ProBEiT
BibTeX
@misc{sychla2026language,
title = {{An RNA Language Model trained on sequence alone reveals the structural logic of Internal Ribosome Entry Sites}},
author = {A. Sychla and P. Bongrand and G. Yang and A. Iyer and J. Rulison and R. A. Wesselhoeft and N. Bisaria and S. Rouskin},
howpublished = {bioRxiv},
year = {2026},
month = may,
doi = {10.64898/2026.05.19.726202},
url = {https://www.biorxiv.org/content/10.64898/2026.05.19.726202v2},
note = {CC BY},
}
RIS
TY - UNPB
TI - An RNA Language Model trained on sequence alone reveals the structural logic of Internal Ribosome Entry Sites
AU - A. Sychla
AU - P. Bongrand
AU - G. Yang
AU - A. Iyer
AU - J. Rulison
AU - R. A. Wesselhoeft
AU - N. Bisaria
AU - S. Rouskin
PY - 2026
DA - 2026/05/19
PB - bioRxiv
DO - 10.64898/2026.05.19.726202
UR - https://www.biorxiv.org/content/10.64898/2026.05.19.726202v2
AB - Millions of RNA sequences are readily available, but the structures that determine their function are not. Picornaviruses initiate translation through Internal Ribosome Entry Sites (IRESes), RNA elements that recruit ribosomes independent of the 5` cap. These elements are large and highly divergent, and structural understanding remains limited to a handful of cases. We address this bottleneck by introducing an RNA language model (Albatross), trained purely on sequence, that predicts high-quality IRES structures at scale. We collect in cellulo chemical probing data for 96 full-length IRESes from divergent viruses and show that Albatross achieves far higher precision (0.80) than state-of-the-art predictions (0.47). Analyzing 75,000 IRES structures, we discover a novel Type II structural subclass and validate it experimentally. We demonstrate the pipeline broadly generalizes to identify functional structures, including tertiary contacts and alternative riboswitch structures. These findings establish Albatross as a scalable framework that accelerates RNA structure discovery and antiviral targeting.
ER -