Commit
·
930c9cf
1
Parent(s):
2abb593
Create README.md
Browse files
README.md
ADDED
@@ -0,0 +1,109 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
licence: unknown
|
3 |
+
---
|
4 |
+
|
5 |
+
# Dataset Card for uracil
|
6 |
+
|
7 |
+
## Table of Contents
|
8 |
+
- [Table of Contents](#table-of-contents)
|
9 |
+
- [Dataset Description](#dataset-description)
|
10 |
+
- [Dataset Summary](#dataset-summary)
|
11 |
+
- [Supported Tasks and Leaderboards](#supported-tasks-and-leaderboards)
|
12 |
+
- [External Use](#external-use)
|
13 |
+
- [PyGeometric](#pygeometric)
|
14 |
+
- [Dataset Structure](#dataset-structure)
|
15 |
+
- [Data Properties](#data-properties)
|
16 |
+
- [Data Fields](#data-fields)
|
17 |
+
- [Data Splits](#data-splits)
|
18 |
+
- [Additional Information](#additional-information)
|
19 |
+
- [Licensing Information](#licensing-information)
|
20 |
+
- [Citation Information](#citation-information)
|
21 |
+
- [Contributions](#contributions)
|
22 |
+
|
23 |
+
## Dataset Description
|
24 |
+
- **[Homepage](http://www.sgdml.org/#datasets)**
|
25 |
+
- **Paper:**: (see citation)
|
26 |
+
|
27 |
+
|
28 |
+
### Dataset Summary
|
29 |
+
The `uracil` dataset is a molecular dynamics (MD) dataset. The total energy and force labels for each dataset were computed using the PBE+vdW-TS electronic structure method. All geometries are in Angstrom, energies and forces are given in kcal/mol and kcal/mol/A respectively.
|
30 |
+
|
31 |
+
|
32 |
+
### Supported Tasks and Leaderboards
|
33 |
+
`uracil` should be used for organic molecular property prediction, a regression task on 1 property. The score used is Mean absolute errors (in meV) for energy prediction.
|
34 |
+
|
35 |
+
|
36 |
+
## External Use
|
37 |
+
### PyGeometric
|
38 |
+
To load in PyGeometric, do the following:
|
39 |
+
|
40 |
+
```python
|
41 |
+
from datasets import load_dataset
|
42 |
+
|
43 |
+
from torch_geometric.data import Data
|
44 |
+
from torch_geometric.loader import DataLoader
|
45 |
+
|
46 |
+
dataset_hf = load_dataset("graphs-datasets/<mydataset>")
|
47 |
+
# For the train set (replace by valid or test as needed)
|
48 |
+
dataset_pg_list = [Data(graph) for graph in dataset_hf["train"]]
|
49 |
+
dataset_pg = DataLoader(dataset_pg_list)
|
50 |
+
```
|
51 |
+
|
52 |
+
## Dataset Structure
|
53 |
+
|
54 |
+
### Data Properties
|
55 |
+
| property | value |
|
56 |
+
|---|---|
|
57 |
+
| scale | big |
|
58 |
+
| #graphs | 133769 |
|
59 |
+
| average #nodes | 12.0 |
|
60 |
+
| average #edges | 128.88676085818943 |
|
61 |
+
|
62 |
+
### Data Fields
|
63 |
+
|
64 |
+
Each row of a given file is a graph, with:
|
65 |
+
- `node_feat` (list: #nodes x #node-features): nodes
|
66 |
+
- `edge_index` (list: 2 x #edges): pairs of nodes constituting edges
|
67 |
+
- `edge_attr` (list: #edges x #edge-features): for the aforementioned edges, contains their features
|
68 |
+
- `y` (list: #labels): contains the number of labels available to predict
|
69 |
+
- `num_nodes` (int): number of nodes of the graph
|
70 |
+
|
71 |
+
### Data Splits
|
72 |
+
|
73 |
+
This data is not split, and should be used with cross validation. It comes from the PyGeometric version of the dataset.
|
74 |
+
|
75 |
+
## Additional Information
|
76 |
+
|
77 |
+
### Licensing Information
|
78 |
+
The dataset has been released under license unknown.
|
79 |
+
|
80 |
+
### Citation Information
|
81 |
+
```
|
82 |
+
@inproceedings{Morris+2020,
|
83 |
+
title={TUDataset: A collection of benchmark datasets for learning with graphs},
|
84 |
+
author={Christopher Morris and Nils M. Kriege and Franka Bause and Kristian Kersting and Petra Mutzel and Marion Neumann},
|
85 |
+
booktitle={ICML 2020 Workshop on Graph Representation Learning and Beyond (GRL+ 2020)},
|
86 |
+
archivePrefix={arXiv},
|
87 |
+
eprint={2007.08663},
|
88 |
+
url={www.graphlearning.io},
|
89 |
+
year={2020}
|
90 |
+
}
|
91 |
+
```
|
92 |
+
|
93 |
+
```
|
94 |
+
|
95 |
+
@article{Chmiela_2017,
|
96 |
+
doi = {10.1126/sciadv.1603015},
|
97 |
+
url = {https://doi.org/10.1126%2Fsciadv.1603015},
|
98 |
+
year = 2017,
|
99 |
+
month = {may},
|
100 |
+
publisher = {American Association for the Advancement of Science ({AAAS})},
|
101 |
+
volume = {3},
|
102 |
+
number = {5},
|
103 |
+
author = {Stefan Chmiela and Alexandre Tkatchenko and Huziel E. Sauceda and Igor Poltavsky and Kristof T. Schütt and Klaus-Robert Müller},
|
104 |
+
title = {Machine learning of accurate energy-conserving molecular force fields},
|
105 |
+
journal = {Science Advances}
|
106 |
+
}
|
107 |
+
|
108 |
+
|
109 |
+
```
|