ZhiyuanChen commited on
Commit
88e4653
·
verified ·
1 Parent(s): 09bad91

Upload folder using huggingface_hub

Browse files
Files changed (2) hide show
  1. README.md +116 -0
  2. test.parquet +3 -0
README.md ADDED
@@ -0,0 +1,116 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ language: rna
3
+ tags:
4
+ - Biology
5
+ - RNA
6
+ - Secondary Structures
7
+ license:
8
+ - cc-by-4.0
9
+ source_datasets:
10
+ - multimolecule/bprna
11
+ - multimolecule/rfam
12
+ task_categories:
13
+ - text-generation
14
+ - fill-mask
15
+ task_ids:
16
+ - language-modeling
17
+ - masked-language-modeling
18
+ pretty_name: IPknot++
19
+ library_name: multimolecule
20
+ ---
21
+
22
+ # IPknot++
23
+
24
+ IPknot++ is a benchmark dataset released with the IPknot++ RNA secondary structure prediction paper.
25
+ It is intended as an external evaluation set rather than a training split.
26
+ A common use is to train or tune RNA secondary structure prediction models on SPOT-RNA-style splits such as [bpRNA-spot](../bprna_spot), then evaluate generalization on IPknot++.
27
+
28
+ The original release does not define train and validation partitions, so every variant is released as a single `test` split.
29
+
30
+ IPknot++ is released as three repositories that share the same RNA targets but differ in whether and how a multiple sequence alignment is provided:
31
+
32
+ - `ipknot_plus_plus`: single-sequence structures with no alignment, covering both the bpRNA-1m and Rfam 14.5 targets.
33
+ - `ipknot_plus_plus-ref`: Rfam 14.5 targets with the Rfam reference alignment, for common-structure prediction.
34
+ - `ipknot_plus_plus-mafft`: Rfam 14.5 targets with a MAFFT alignment, for common-structure prediction.
35
+
36
+ The `-ref` and `-mafft` variants cover the same set of Rfam 14.5 targets, with identical `id`, `sequence`, and `secondary_structure`; only `aligned_sequences` differs (the alignment members in `aligned_ids` are also identical). These targets are a subset of the single sequences in the no-alignment `ipknot_plus_plus` variant.
37
+
38
+ For the alignment variants, `sequence` and `secondary_structure` are taken from the corresponding single-sequence BPSEQ file, and the full alignment is stored in `aligned_ids` and `aligned_sequences`.
39
+
40
+ ## Schema
41
+
42
+ The `ipknot_plus_plus` variant contains:
43
+
44
+ | Column | Description |
45
+ | --- | --- |
46
+ | `id` | Identifier of the benchmark entry. |
47
+ | `sequence` | Ungapped RNA sequence for the benchmark target. |
48
+ | `secondary_structure` | Target secondary structure in dot-bracket notation. |
49
+
50
+ The `ipknot_plus_plus-ref` and `ipknot_plus_plus-mafft` variants add two alignment columns:
51
+
52
+ | Column | Description |
53
+ | --- | --- |
54
+ | `aligned_ids` | Sequence IDs in the alignment. The first entry is the benchmark target itself. |
55
+ | `aligned_sequences` | Aligned sequences, preserving gap characters. |
56
+
57
+ ## Disclaimer
58
+
59
+ This is an UNOFFICIAL release of the IPknot++ benchmark dataset by Kengo Sato and Yuki Kato.
60
+
61
+ **The team releasing IPknot++ did not write this dataset card for this dataset so this dataset card has been written by the MultiMolecule team.**
62
+
63
+ ## Dataset Description
64
+
65
+ - **Homepage**: https://www.sato-lab.org/en/publication/sato-2022-qk/
66
+ - **datasets**: https://huggingface.co/datasets/multimolecule/ipknot_plus_plus
67
+ - **Original URL**: https://zenodo.org/records/4923158
68
+
69
+ ## License
70
+
71
+ The original IPknot++ benchmark dataset is licensed under the [Creative Commons Attribution 4.0 International](https://creativecommons.org/licenses/by/4.0/) license.
72
+
73
+ ```spdx
74
+ SPDX-License-Identifier: CC-BY-4.0
75
+ ```
76
+
77
+ ## Citation
78
+
79
+ ```bibtex
80
+ @article{sato2022prediction,
81
+ author = {Sato, Kengo and Kato, Yuki},
82
+ journal = {Briefings in Bioinformatics},
83
+ month = jan,
84
+ number = 1,
85
+ title = {Prediction of {RNA} secondary structure including pseudoknots for long sequences},
86
+ volume = 23,
87
+ year = 2022
88
+ }
89
+ ```
90
+
91
+ ```bibtex
92
+ @dataset{sato2021ipknot,
93
+ author = {Sato, Kengo and Kato, Yuki},
94
+ title = {{IPknot++} benchmark dataset},
95
+ publisher = {Zenodo},
96
+ doi = {10.5281/zenodo.4923158},
97
+ year = 2021
98
+ }
99
+ ```
100
+
101
+ > [!NOTE]
102
+ > The artifacts distributed in this repository are part of the MultiMolecule project.
103
+ > If MultiMolecule supports your research, please cite the MultiMolecule project as follows:
104
+
105
+ ```bibtex
106
+ @software{chen_2024_12638419,
107
+ author = {Chen, Zhiyuan and Zhu, Sophia Y.},
108
+ title = {MultiMolecule},
109
+ doi = {10.5281/zenodo.12638419},
110
+ publisher = {Zenodo},
111
+ url = {https://doi.org/10.5281/zenodo.12638419},
112
+ year = 2024,
113
+ month = may,
114
+ day = 4
115
+ }
116
+ ```
test.parquet ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:62a04bdc412c84926feeb7c09db4af50bdd7ba39ba05132728dd58a52502f851
3
+ size 905512