phosseini commited on
Commit
a1d5fe6
·
verified ·
1 Parent(s): 4afa948

Rename to oligogym; add OligoGym attribution, citation, full source lists

Browse files
Files changed (1) hide show
  1. README.md +100 -69
README.md CHANGED
@@ -1,84 +1,95 @@
1
  ---
2
  license: cc-by-4.0
3
  task_categories:
4
- - tabular-regression
5
  tags:
6
- - biology
7
- - oligonucleotides
8
- - siRNA
9
- - ASO
10
- - shRNA
11
- - HELM
 
12
  configs:
13
- - config_name: alharbi_2020_1
14
- data_files:
15
- - split: train
16
- path: alharbi_2020_1.csv.gz
17
- - config_name: alharbi_2020_2
18
- data_files:
19
- - split: train
20
- path: alharbi_2020_2.csv.gz
21
- - config_name: hagedorn_2022_1
22
- data_files:
23
- - split: train
24
- path: hagedorn_2022_1.csv.gz
25
- - config_name: hwang_2024_1
26
- data_files:
27
- - split: train
28
- path: hwang_2024_1.csv.gz
29
- - config_name: ichihara_2007_1
30
- data_files:
31
- - split: train
32
- path: ichihara_2007_1.csv.gz
33
- - config_name: ichihara_2007_2
34
- data_files:
35
- - split: train
36
- path: ichihara_2007_2.csv.gz
37
- - config_name: knott_2014_1
38
- data_files:
39
- - split: train
40
- path: knott_2014_1.csv.gz
41
- - config_name: martinelli_2023_1
42
- data_files:
43
- - split: train
44
- path: martinelli_2023_1.csv.gz
45
- - config_name: mcquisten_2007_1
46
- data_files:
47
- - split: train
48
- path: mcquisten_2007_1.csv.gz
49
- - config_name: moe_neurotox_1
50
- data_files:
51
- - split: train
52
- path: moe_neurotox_1.csv.gz
53
- - config_name: papargyri_2020_1
54
- data_files:
55
- - split: train
56
- path: papargyri_2020_1.csv.gz
57
- - config_name: shmushkovich_2018_1
58
- data_files:
59
- - split: train
60
- path: shmushkovich_2018_1.csv.gz
61
- - config_name: metadata
62
- data_files:
63
- - split: train
64
- path: metadata.csv
65
  ---
66
 
67
- # CollageBio Oligonucleotide Datasets
 
 
 
 
 
 
 
 
 
 
68
 
69
- Curated benchmark datasets for oligonucleotide activity, toxicity, and
70
- immunomodulation prediction. Each dataset is stored as a gzip-compressed
71
- CSV with columns:
72
 
73
  | Column | Description |
74
  |---|---|
75
  | `x` | HELM string for the oligonucleotide |
76
- | `y` | Numeric label (activity / toxicity score) |
 
77
  | `targets` | mRNA target identifier |
 
 
 
 
78
 
79
  ## Datasets
80
 
81
- | Key | Canonical Name | OligoGym ID | Modality | Collection | Samples | Source |
82
  |---|---|---|---|---|---|---|
83
  | `ichihara_2007_1` | Ichihara_2007_1 | Huesken | SIRNA | activity | 2,431 | [DOI](https://doi.org/10.1093/nar/gkm699) |
84
  | `ichihara_2007_2` | Ichihara_2007_2 | Ichihara | SIRNA | activity | 419 | [DOI](https://doi.org/10.1093/nar/gkm699) |
@@ -87,21 +98,41 @@ CSV with columns:
87
  | `hagedorn_2022_1` | Hagedorn_2022_1 | Neurotox LNA | ASO | toxicity | 1,825 | [DOI](https://doi.org/10.1089/nat.2021.0071) |
88
  | `hwang_2024_1` | Hwang_2024_1 | ASOptimizer | ASO | activity | 32,602 | [DOI](https://doi.org/10.1016/j.omtn.2024.102186) |
89
  | `knott_2014_1` | Knott_2014_1 | Sherwood | SHRNA | activity | 291,551 | [DOI](https://doi.org/10.1016/j.molcel.2014.10.025) |
90
- | `moe_neurotox_1` | MOE_Neurotox_1 | Neurotox MOE | ASO | neurotoxicity | 2,437 | [DOI](https://patents.google.com/patent/WO2020172559A1) |
91
  | `martinelli_2023_1` | Martinelli_2023_1 | siRNAmod | SIRNA | activity | 907 | [DOI](https://doi.org/10.1016/j.ygeno.2024.110815) |
92
- | `mcquisten_2007_1` | McQuisten_2007_1 | OpenASO | ASO | activity | 3,913 | [DOI](https://doi.org/10.1186/1471-2105-8-184) |
93
  | `papargyri_2020_1` | Papargyri_2020_1 | Cytotox LNA | ASO | toxicity | 768 | [DOI](https://doi.org/10.1016/j.omtn.2019.12.011) |
94
  | `shmushkovich_2018_1` | Shmushkovich_2018_1 | Shmushkovich | SIRNA | activity | 356 | [DOI](https://doi.org/10.1093/nar/gky745) |
95
 
 
 
 
 
96
  ## Usage
97
 
98
  ```python
99
  from datasets import load_dataset
100
 
101
  # Load a single dataset
102
- ds = load_dataset("CollageBio/oligo-datasets", name="ichihara_2007_1")
103
  df = ds["train"].to_pandas()
104
 
105
  # Or download the raw file directly
106
- # https://huggingface.co/datasets/CollageBio/oligo-datasets/resolve/main/ichihara_2007_1.csv.gz
107
  ```
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
  ---
2
  license: cc-by-4.0
3
  task_categories:
4
+ - tabular-regression
5
  tags:
6
+ - biology
7
+ - oligonucleotides
8
+ - siRNA
9
+ - ASO
10
+ - shRNA
11
+ - HELM
12
+ - oligogym
13
  configs:
14
+ - config_name: ichihara_2007_1
15
+ data_files:
16
+ - split: train
17
+ path: ichihara_2007_1.csv.gz
18
+ - config_name: ichihara_2007_2
19
+ data_files:
20
+ - split: train
21
+ path: ichihara_2007_2.csv.gz
22
+ - config_name: alharbi_2020_1
23
+ data_files:
24
+ - split: train
25
+ path: alharbi_2020_1.csv.gz
26
+ - config_name: alharbi_2020_2
27
+ data_files:
28
+ - split: train
29
+ path: alharbi_2020_2.csv.gz
30
+ - config_name: hagedorn_2022_1
31
+ data_files:
32
+ - split: train
33
+ path: hagedorn_2022_1.csv.gz
34
+ - config_name: hwang_2024_1
35
+ data_files:
36
+ - split: train
37
+ path: hwang_2024_1.csv.gz
38
+ - config_name: knott_2014_1
39
+ data_files:
40
+ - split: train
41
+ path: knott_2014_1.csv.gz
42
+ - config_name: moe_neurotox_1
43
+ data_files:
44
+ - split: train
45
+ path: moe_neurotox_1.csv.gz
46
+ - config_name: martinelli_2023_1
47
+ data_files:
48
+ - split: train
49
+ path: martinelli_2023_1.csv.gz
50
+ - config_name: mcquisten_2007_1
51
+ data_files:
52
+ - split: train
53
+ path: mcquisten_2007_1.csv.gz
54
+ - config_name: papargyri_2020_1
55
+ data_files:
56
+ - split: train
57
+ path: papargyri_2020_1.csv.gz
58
+ - config_name: shmushkovich_2018_1
59
+ data_files:
60
+ - split: train
61
+ path: shmushkovich_2018_1.csv.gz
62
+
 
 
 
63
  ---
64
 
65
+ # CollageBio OligoGym Datasets
66
+
67
+ A mirror of the 12 curated benchmark datasets shipped with
68
+ [**OligoGym**](https://github.com/Roche/OligoGym) (Roche), repackaged as HuggingFace configs for
69
+ convenient loading. Covers oligonucleotide activity, toxicity, and
70
+ immunomodulation across ASO, siRNA, and shRNA modalities.
71
+
72
+ > **This is not a new dataset.** Row counts, column sets, and values are
73
+ > identical to the files in OligoGym's `oligogym/resources/pkg_dataset/`.
74
+ > All credit for curation belongs to the OligoGym authors — please cite them
75
+ > (see [Citation](#citation)) **and** each dataset's own primary source.
76
 
77
+ ## Columns
 
 
78
 
79
  | Column | Description |
80
  |---|---|
81
  | `x` | HELM string for the oligonucleotide |
82
+ | `y` | Numeric label, as processed by OligoGym (the modelling target) |
83
+ | `y_raw` | Original label before clipping / scaling — differs from `y` for `hwang_2024_1`, `knott_2014_1`, `moe_neurotox_1`, `papargyri_2020_1` |
84
  | `targets` | mRNA target identifier |
85
+ | `smiles` | SMILES representation |
86
+ | `fasta` | Nucleotide sequence |
87
+
88
+ `hwang_2024_1` additionally carries `cell_line` and `concentration`.
89
 
90
  ## Datasets
91
 
92
+ | Key | Canonical Name | OligoGym ID | Modality | Collection | Samples | Source(s) |
93
  |---|---|---|---|---|---|---|
94
  | `ichihara_2007_1` | Ichihara_2007_1 | Huesken | SIRNA | activity | 2,431 | [DOI](https://doi.org/10.1093/nar/gkm699) |
95
  | `ichihara_2007_2` | Ichihara_2007_2 | Ichihara | SIRNA | activity | 419 | [DOI](https://doi.org/10.1093/nar/gkm699) |
 
98
  | `hagedorn_2022_1` | Hagedorn_2022_1 | Neurotox LNA | ASO | toxicity | 1,825 | [DOI](https://doi.org/10.1089/nat.2021.0071) |
99
  | `hwang_2024_1` | Hwang_2024_1 | ASOptimizer | ASO | activity | 32,602 | [DOI](https://doi.org/10.1016/j.omtn.2024.102186) |
100
  | `knott_2014_1` | Knott_2014_1 | Sherwood | SHRNA | activity | 291,551 | [DOI](https://doi.org/10.1016/j.molcel.2014.10.025) |
101
+ | `moe_neurotox_1` | MOE_Neurotox_1 | Neurotox MOE | ASO | acute neurotoxicity | 2,437 | [1](https://patents.google.com/patent/WO2020172559A1) [2](https://patents.google.com/patent/WO2016168592A2) [3](https://patents.google.com/patent/WO2021021673A1) [4](https://patents.google.com/patent/WO2015153800A2) [5](https://patents.google.com/patent/WO2020205463A1) [6](https://patents.google.com/patent/WO2020006267A1) [7](https://patents.google.com/patent/WO2020023737A1) [8](https://patents.google.com/patent/WO2020243292A1) [9](https://patents.google.com/patent/WO2020106996A1) [10](https://patents.google.com/patent/WO2022026589A1) [11](https://patents.google.com/patent/WO2022165122A1) [12](https://patents.google.com/patent/WO2021174019A1) [13](https://patents.google.com/patent/WO2022032060A2) |
102
  | `martinelli_2023_1` | Martinelli_2023_1 | siRNAmod | SIRNA | activity | 907 | [DOI](https://doi.org/10.1016/j.ygeno.2024.110815) |
103
+ | `mcquisten_2007_1` | McQuisten_2007_1 | OpenASO | ASO | activity | 3,913 | [1](https://doi.org/10.1186/1471-2105-8-184) [2](https://github.com/lackeylela/openASO) |
104
  | `papargyri_2020_1` | Papargyri_2020_1 | Cytotox LNA | ASO | toxicity | 768 | [DOI](https://doi.org/10.1016/j.omtn.2019.12.011) |
105
  | `shmushkovich_2018_1` | Shmushkovich_2018_1 | Shmushkovich | SIRNA | activity | 356 | [DOI](https://doi.org/10.1093/nar/gky745) |
106
 
107
+ `moe_neurotox_1` is compiled from 13 separate WO (PCT) patent publications;
108
+ `mcquisten_2007_1` from a publication plus the openASO repository. All sources
109
+ are linked above.
110
+
111
  ## Usage
112
 
113
  ```python
114
  from datasets import load_dataset
115
 
116
  # Load a single dataset
117
+ ds = load_dataset("CollageBio/oligogym", name="ichihara_2007_1")
118
  df = ds["train"].to_pandas()
119
 
120
  # Or download the raw file directly
121
+ # https://huggingface.co/datasets/CollageBio/oligogym/resolve/main/ichihara_2007_1.csv.gz
122
  ```
123
+
124
+ ## Citation
125
+
126
+ If you use these datasets, please cite the OligoGym paper:
127
+
128
+ ```bibtex
129
+ @article{rotrattanadumrong2026oligogym,
130
+ title={OligoGym: Curated Datasets and Benchmarks for Oligonucleotide Drug Discovery},
131
+ author={Rotrattanadumrong, Rachapun and De Donno, Carlo},
132
+ journal={Advances in Neural Information Processing Systems},
133
+ volume={38},
134
+ year={2026}
135
+ }
136
+ ```
137
+
138
+ Upstream code and original data: [https://github.com/Roche/OligoGym](https://github.com/Roche/OligoGym)