Datasets:

multimolecule
/

chanrg

language: rna
tags:
  - Biology
  - RNA
license:
  - agpl-3.0
size_categories:
  - 100K<n<1M
source_datasets:
  - multimolecule/rfam
task_categories:
  - text-generation
  - fill-mask
task_ids:
  - language-modeling
  - masked-language-modeling
pretty_name: CHANRG
library_name: multimolecule

Comprehensive Hierarchical Annotation of Non-coding RNA Groups (CHANRG)

CHANRG is a database of non-coding RNA families and secondary structures.

Dataset Description

Homepage: https://multimolecule.danling.org/datasets/chanrg
datasets: https://huggingface.co/datasets/multimolecule/chanrg
Point of Contact: Zhiyuan Chen

Example Entry

id	sequence	secondary_structure	structural_annotation	functional_annotation	family	clan	architecture	super_family	split
AAAA02037454.1_2001-2135	GGATGCGATCATACCAGCACTAAAGCACCGGA...	(((((((((....((.(((((...((..((((...	SSSSSSSSSMMMMSSISSSSSIIISSIISSSS...	NNNNNNNNNNNNNNNNNNNNNNNNNNNNNNNN...	RF00001	CL00113	rRNA	5S_rRNA	Train

Column Description

The converted dataset consists of the following columns, each providing specific information about the RNA secondary structures, consistent with the bpRNA standard:

id: A unique identifier for each RNA entry. This ID is derived from the original .sta file name and serves as a reference to the specific RNA structure within the dataset.
sequence: The nucleotide sequence of the RNA molecule, represented using the standard RNA bases:
- A: Adenine
- C: Cytosine
- G: Guanine
- U: Uracil
secondary_structure: The secondary structure of the RNA represented in dot-bracket notation, using up to three types of symbols to indicate base pairing and unpaired regions, as per bpRNA's standard:
- Dots (.): Represent unpaired nucleotides.
- Parentheses (( and )): Represent base pairs in standard stems (page 1).
- Square Brackets ([ and ]): Represent base pairs in pseudoknots (page 2).
- Curly Braces ({ and }): Represent base pairs in additional pseudoknots (page 3).
structural_annotation: Structural annotations categorizing different regions of the RNA based on their roles within the secondary structure, consistent with bpRNA standards:
- E: External Loop – Regions that are unpaired and external to any loop or helix.
- S: Stem – Paired regions forming helical structures.
- H: Hairpin Loop – Unpaired regions at the end of a stem, forming a loop.
- I: Internal Loop – Unpaired regions between two stems.
- M: Multi-loop – Junctions where three or more stems converge.
- B: Bulge – Unpaired nucleotides on one side of a stem.
- X: Ambiguous or Undetermined – Regions where the structure is unknown or cannot be classified.
- K: Pseudoknot – Regions involved in pseudoknots, where base pairs cross each other.
functional_annotation: Functional annotations indicating specific functional elements or regions within the RNA sequence, as defined by bpRNA:
- N: None – No specific functional annotation is assigned.
- K: Pseudoknot – Marks nucleotides involved in pseudoknot structures, which can be functionally significant.
family: The Rfam family accession for the RNA entry, such as RF00001. Entries with the same family belong to the same Rfam family model and share the same family-level annotation.
clan: The Rfam clan accession for the family, such as CL00113. Clans group related Rfam families into a broader category. This field can be None when a family has no clan assignment in Rfam.

The architecture and super_family fields follow the RNArchitecture hierarchy described by Boccaletto et al. (2018).

architecture: A coarse-grained structural architecture label following the hierarchical RNA classification described in the RNArchitecture paper, such as rRNA, hairpin, 3WJ, or complex unclassified. This field captures the high-level structural organization shared by structurally similar RNA families.
super_family: The super-family label following the same RNArchitecture classification, such as 5S_rRNA. It groups evolutionarily related RNA families within the broader architecture. This field can be None when no super-family annotation is available.
split: The CHANRG split label assigned during dataset conversion.
- Train: Used for model training.
- Validation: Held out for model selection.
- Test: Standard held-out evaluation set.
- GenF: Entire families held out because they cannot be split without accession leakage, used for family-level generalization evaluation.
- GenC: Families from clans without a super-family assignment, held out for clan-level generalization evaluation.
- GenA: Families with architecture equal to complex unclassified, held out for architecture-level generalization evaluation.

License

This dataset is licensed under the GNU Affero General Public License.

For additional terms and clarifications, please refer to our License FAQ.

SPDX-License-Identifier: AGPL-3.0-or-later

Citation

@article{chen2026fair,
  title         = "Fair splits flip the leaderboard: {CHANRG} reveals limited generalization in {RNA} secondary-structure prediction",
  author        = "Chen, Zhiyuan and Deng, Zhenfeng and Deng, Pan and Liao, Yue and Su, Xiu and Ye, Peng and Liu, Xihui",
  month         =  mar,
  year          =  2026,
  copyright     = "http://creativecommons.org/licenses/by-nc-sa/4.0/",
  archivePrefix = "arXiv",
  primaryClass  = "q-bio.BM",
  eprint        = "2603.22330"
}

The artifacts distributed in this repository are part of the MultiMolecule project. If you use MultiMolecule in your research, you must cite the MultiMolecule project as follows:

@software{chen_2024_12638419,
  author    = {Chen, Zhiyuan and Zhu, Sophia Y.},
  title     = {MultiMolecule},
  doi       = {10.5281/zenodo.12638419},
  publisher = {Zenodo},
  url       = {https://doi.org/10.5281/zenodo.12638419},
  year      = 2024,
  month     = may,
  day       = 4
}