haneulpark commited on
Commit
1f37c2e
·
verified ·
1 Parent(s): 292dfdb

Upload README.md with huggingface_hub

Browse files
Files changed (1) hide show
  1. README.md +153 -167
README.md CHANGED
@@ -1,167 +1,153 @@
1
- ---
2
- license: mit
3
- dataset_info:
4
- - config_name: ADME
5
- features:
6
- - name: Task
7
- dtype: string
8
- - name: Drug_ID
9
- dtype: string
10
- - name: SMILES
11
- dtype: string
12
- - name: Y
13
- dtype: float64
14
- - name: split
15
- dtype: string
16
- splits:
17
- - name: train
18
- num_bytes: 11857638
19
- num_examples: 111612
20
- download_size: 4050495
21
- dataset_size: 11857638
22
- - config_name: CRISPROutcome
23
- features:
24
- - name: Task
25
- dtype: string
26
- - name: GuideSeq_ID
27
- dtype: string
28
- - name: GuideSeq
29
- dtype: string
30
- - name: Y
31
- dtype: float64
32
- - name: split
33
- dtype: string
34
- splits:
35
- - name: train
36
- num_bytes: 681958
37
- num_examples: 7605
38
- download_size: 215266
39
- dataset_size: 681958
40
- - config_name: Develop
41
- features:
42
- - name: Task
43
- dtype: string
44
- - name: Antibody_ID
45
- dtype: string
46
- - name: heavy_chain
47
- dtype: string
48
- - name: light_chain
49
- dtype: string
50
- - name: Y
51
- dtype: float64
52
- - name: split
53
- dtype: string
54
- splits:
55
- - name: train
56
- num_bytes: 732537
57
- num_examples: 3614
58
- download_size: 235181
59
- dataset_size: 732537
60
- - config_name: Epitope
61
- features:
62
- - name: Task
63
- dtype: string
64
- - name: Antigen_ID
65
- dtype: string
66
- - name: Antigen
67
- dtype: string
68
- - name: Y
69
- dtype: string
70
- - name: split
71
- dtype: string
72
- splits:
73
- - name: train
74
- num_bytes: 2504514
75
- num_examples: 3606
76
- download_size: 2179200
77
- dataset_size: 2504514
78
- - config_name: HTS
79
- features:
80
- - name: Task
81
- dtype: string
82
- - name: Drug_ID
83
- dtype: string
84
- - name: SMILES
85
- dtype: string
86
- - name: Y
87
- dtype: int64
88
- - name: split
89
- dtype: string
90
- splits:
91
- - name: train
92
- num_bytes: 234479816
93
- num_examples: 2053052
94
- download_size: 58513481
95
- dataset_size: 234479816
96
- - config_name: QM
97
- features:
98
- - name: Task
99
- dtype: string
100
- - name: Drug_ID
101
- dtype: string
102
- - name: Atom
103
- dtype: string
104
- - name: x_coordinate
105
- dtype: float64
106
- - name: y_coordinate
107
- dtype: float64
108
- - name: z_coordinate
109
- dtype: float64
110
- - name: Y
111
- dtype: float64
112
- - name: split
113
- dtype: string
114
- splits:
115
- - name: train
116
- num_bytes: 2896357941
117
- num_examples: 41522485
118
- download_size: 1249890104
119
- dataset_size: 2896357941
120
- - config_name: Tox
121
- features:
122
- - name: Task
123
- dtype: string
124
- - name: Drug_ID
125
- dtype: string
126
- - name: SMILES
127
- dtype: string
128
- - name: Y
129
- dtype: float64
130
- - name: split
131
- dtype: string
132
- splits:
133
- - name: train
134
- num_bytes: 258380329
135
- num_examples: 2563685
136
- download_size: 78889561
137
- dataset_size: 258380329
138
- configs:
139
- - config_name: ADME
140
- data_files:
141
- - split: train
142
- path: single_instance_datasets/ADME/train-*
143
- - config_name: CRISPROutcome
144
- data_files:
145
- - split: train
146
- path: single_instance_datasets/CRISPROutcome/train-*
147
- - config_name: Develop
148
- data_files:
149
- - split: train
150
- path: single_instance_datasets/Develop/train-*
151
- - config_name: Epitope
152
- data_files:
153
- - split: train
154
- path: single_instance_datasets/Epitope/train-*
155
- - config_name: HTS
156
- data_files:
157
- - split: train
158
- path: single_instance_datasets/HTS/train-*
159
- - config_name: QM
160
- data_files:
161
- - split: train
162
- path: single_instance_datasets/QM/train-*
163
- - config_name: Tox
164
- data_files:
165
- - split: train
166
- path: single_instance_datasets/Tox/train-*
167
- ---
 
1
+ ---
2
+ verison: 1.0.0
3
+ license: mit
4
+ task_categories:
5
+ - tabular-classification
6
+ - tabular-regression
7
+ language:
8
+ - en
9
+ tags:
10
+ - bioactivity
11
+ - therapeutic science
12
+ pretty_name: Therapeutics Data Commons
13
+ size_categories:
14
+ - 10M<n<100M
15
+ dataset_summary: >-
16
+ Therapeutics Data Commons (TDC) provides curated, AI-ready datasets, machine
17
+ learning tasks, and benchmarks with meaningful data splits, supporting the
18
+ development and evaluation of AI methods for therapeutic discovery. TDC tasks
19
+ are categorized into three main problem types: (1) single-instance prediction,
20
+ (2) multi-instance learning, and (3) molecule generation.
21
+ citation: >-
22
+ @article{Huang2021tdc,
23
+ title={Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development},
24
+ author={Huang, Kexin and Fu, Tianfan and Gao, Wenhao and Zhao, Yue and Roohani, Yusuf and Leskovec, Jure and Coley, Connor W and Xiao, Cao and Sun, Jimeng and Zitnik, Marinka},
25
+ journal={Proceedings of Neural Information Processing Systems, NeurIPS Datasets and Benchmarks},
26
+ year={2021}
27
+ }
28
+
29
+ @article{Huang2022artificial,
30
+ title={Artificial intelligence foundation for therapeutic science},
31
+ author={Huang, Kexin and Fu, Tianfan and Gao, Wenhao and Zhao, Yue and Roohani, Yusuf and Leskovec, Jure and Coley, Connor W and Xiao, Cao and Sun, Jimeng and Zitnik, Marinka},
32
+ journal={Nature Chemical Biology},
33
+ year={2022}
34
+ }
35
+
36
+ @article{velez-arce2024signals,
37
+ title={Signals in the Cells: Multimodal and Contextualized Machine Learning Foundations for Therapeutics},
38
+ author={Velez-Arce, Alejandro and Lin, Xiang and Huang, Kexin and Li, Michelle M and Gao, Wenhao and Pentelute, Bradley and Fu, Tianfan and Kellis, Manolis and Zitnik, Marinka},
39
+ booktitle={NeurIPS 2024 Workshop on AI for New Drug Modalities},
40
+ year={2024}
41
+ }
42
+
43
+ # Single instance prediciton
44
+ config_names:
45
+ - ADME
46
+ - Tox
47
+ - HTS
48
+ - CRISPROutcome
49
+ - Develop
50
+ - Epitope
51
+ - QM
52
+ configs:
53
+ - config_name: ADME
54
+ data_files: single_instance_prediction_datasets/ADME.parquet
55
+ columns:
56
+ - Task
57
+ - Drug_ID
58
+ - SMILES
59
+ - 'Y'
60
+ - split
61
+ - config_name: Tox
62
+ data_files: single_instance_prediction_datasets/Tox.parquet
63
+ columns:
64
+ - Task
65
+ - Drug_ID
66
+ - SMILES
67
+ - 'Y'
68
+ - split
69
+ - config_name: HTS
70
+ data_files: single_instance_prediction_datasets/HTS.parquet
71
+ columns:
72
+ - Task
73
+ - Drug_ID
74
+ - SMILES
75
+ - 'Y'
76
+ - split
77
+ - config_name: CRISPROutcome
78
+ data_files: single_instance_prediction_datasets/CRISPROutcome.parquet
79
+ columns:
80
+ - Task
81
+ - GuideSeq_ID
82
+ - GuideSeq
83
+ - 'Y'
84
+ - split
85
+ - config_name: Develop
86
+ data_files: single_instance_prediction_datasets/Develop.parquet
87
+ columns:
88
+ - Task
89
+ - Antibody_ID
90
+ - heavy_chain
91
+ - light_chain
92
+ - 'Y'
93
+ - split
94
+ - config_name: Epitope
95
+ data_files: single_instance_prediction_datasets/Epitope.parquet
96
+ columns:
97
+ - Task
98
+ - Antigen_ID
99
+ - Antigen
100
+ - 'Y'
101
+ - split
102
+ - config_name: QM
103
+ data_files: single_instance_prediction_datasets/QM.parquet
104
+ columns:
105
+ - Task
106
+ - Drug_ID
107
+ - Atom
108
+ - x_coordinate
109
+ - y_coordinate
110
+ - z_coordinate
111
+ - 'Y'
112
+ - split
113
+
114
+
115
+ ---
116
+ # Therapeutics Data Commons
117
+ [Therapeutics Data Commons](https://tdcommons.ai/)(TDC) provides a publicly available collection of 22 machine learning tasks for therapeutic discovery. Our Hugging Face repository is a mirror of single-instance prediction tasks of TDC, encompassing a total of 46,265,659 data points.
118
+
119
+ The parquet files uploaded to our Hugging Face repository have been sanitized and curated from the original datasets.
120
+ - Each parquet file corresponds to a separate category (e.g., ADME) and contains multiple tasks (e.g., solubility, permeability).
121
+ - Each file follows its own schema (i.e., different column names) and has been preprocessed differently depending on the category.
122
+ - As ADME, Tox, and HTS datasets contain SMILES strings, we have sanitized (standardized) the SMILES strings. The sanitization process includes removing salts, standardizing the SMILES strings to a canonical form, etc. 2 invalid SMILES strings from ADME dataset and 59 invalid SMILES strings from HTS dataset were removed during the sanitization process.
123
+
124
+ A summary file (i.e., single_instance_prediction_summary.csv) is also uploaded, which lists:
125
+ - Problem (Category),
126
+ - Task,
127
+ - DatasetName,
128
+ - NumCompounds,
129
+ - Leaderboard information (Unit, Task, Metric, Dataset Split - https://tdcommons.ai/benchmark/admet_group/overview/)
130
+ - License,
131
+ - DatasetDescription,
132
+ - TaskDescription,
133
+ - Reference
134
+
135
+
136
+
137
+ ## Quick Usage
138
+ Load a dataset in python
139
+
140
+ Each subset can be loaded into python using the Huggingface [datasets](https://huggingface.co/docs/datasets/index) library.
141
+ First, from the command line install the `datasets` library
142
+
143
+ $ pip install datasets
144
+
145
+ then, from within python load the datasets library.
146
+
147
+ >>> import datasets
148
+
149
+ Now load one of the 'TDC' datasets, e.g.,
150
+
151
+ >>> dataset = datasets.load_dataset("maomlab/TDC", name = "ADME")
152
+
153
+ You can modify "name" based on your interest (e.g., "Tox").