README.md

metadata

license: unknown
tags:
  - biology
  - rna
  - genomics

Overview

Internal ribosome entry sites (IRESs) enable cap-independent translation initiation. This binary classification dataset contains 46,774 unique sequences: 37,602 non-IRES examples and 9,172 IRES examples. The target corresponds to the source dataset's IRES_class_600 label.

The sequences combine a high-throughput oligo assay with entries from IRESbase, IRESite, Rfam, and the IRESpred training data. The original source collection is retained for each row. The dataset mixes 5' UTR fragments with viral-genome/CDS fragments, known IRES mutants, synthetic constructs, and controls.

This dataset is redistributed as part of mRNABench: https://github.com/morrislab/mRNABench

Data Format

Description of data columns:

• transcript_id: Identifier supplied by the source dataset.
• sequence: Candidate IRES sequence using the DNA alphabet.
• cds: Length-matched all-zero track because consistent coding-sequence annotations are unavailable.
• splice: Length-matched all-zero track because consistent splice-junction annotations are unavailable.
• target: Binary IRES activity label, where 1 is IRES and 0 is non-IRES.
• description: Original sequence collection in source=<name> format.

Data Source

This dataset is a Hugging Face redistribution of the IRES classification dataset assembled for UTR-LM. It combines multiple source collections with different licensing terms, and no single redistribution license is specified. Listed below are the data sources and citation:

Original paper: https://doi.org/10.1038/s42256-024-00823-9
Original dataset source: https://github.com/a96123155/IRES_Prediction_Design
Citation: Chu, Y. et al. A 5' UTR language model for decoding untranslated regions of mRNA and function predictions. Nat Mach Intell 6, 449-460 (2024). https://doi.org/10.1038/s42256-024-00823-9