Datasets:
Update README.md
Browse files
README.md
CHANGED
@@ -1,3 +1,124 @@
|
|
1 |
-
---
|
2 |
-
license: cdla-permissive-2.0
|
3 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
1 |
+
---
|
2 |
+
license: cdla-permissive-2.0
|
3 |
+
tags:
|
4 |
+
- chemistry
|
5 |
+
- smiles
|
6 |
+
- cheminformatics
|
7 |
+
pretty_name: MSR-ACC TAE25 Regression
|
8 |
+
size_categories:
|
9 |
+
- 10K<n<100K
|
10 |
+
---
|
11 |
+
|
12 |
+
# MSR-ACC TAE25 Regression Dataset
|
13 |
+
|
14 |
+
This dataset is a processed version of the Microsoft Research Accurate Chemistry Collection (MSR-ACC) TAE25 dataset, containing thermochemical data for molecules with computed Total Atomization Energies (TAE) using various quantum chemical methods.
|
15 |
+
|
16 |
+
[](https://doi.org/10.5281/zenodo.15783826)
|
17 |
+
|
18 |
+
## Dataset Description
|
19 |
+
|
20 |
+
The original MSR-ACC TAE25 dataset contains QCSchema JSON files with 3D molecular geometries and computed thermochemical properties.
|
21 |
+
|
22 |
+
This processed version provides:
|
23 |
+
- **SMILES representations** generated from 3D coordinates and graph connectivity
|
24 |
+
- **Train/validation/test splits** for machine learning applications
|
25 |
+
- **Train set**: Contains molecular structures from the original training partition (except test samples)
|
26 |
+
- **Validation set**: Contains molecular structures from the original validation partition
|
27 |
+
- **Test set**: 1,510 molecules sampled from the training set (1,485 random samples + 25 complete records)
|
28 |
+
|
29 |
+
## Data Processing Pipeline
|
30 |
+
|
31 |
+
### 1. SMILES Generation
|
32 |
+
|
33 |
+
SMILES strings were generated from 3D molecular coordinates using the following process:
|
34 |
+
|
35 |
+
1. **QCSchema Parsing**: Loaded QCSchema JSON files using `qcelemental` for validation
|
36 |
+
2. **Bonds (Graph Connectivity)**: Used pre-computed connectivity from `graph:all` fields in QCSchema extras
|
37 |
+
3. **SMILES Generation**: Generated canonical, isomeric SMILES using RDKit with `MolToSmiles(canonical=True, isomericSmiles=True)`
|
38 |
+
|
39 |
+
### 2. Dataset Splitting
|
40 |
+
|
41 |
+
The test set was created by sampling from the original training partition to ensure proper evaluation:
|
42 |
+
|
43 |
+
1. **Sampling**:
|
44 |
+
- Selected 25 molecules with complete (non-NaN) target values across all thermochemical properties
|
45 |
+
- Randomly sampled 1,485 additional molecules from remaining training data
|
46 |
+
|
47 |
+
2. **Data Cleaning**:
|
48 |
+
- Maintained original train/validation split from source dataset
|
49 |
+
|
50 |
+
## Data Format
|
51 |
+
|
52 |
+
Each Parquet file contains the following columns:
|
53 |
+
|
54 |
+
- `filename`: QCSchema JSON filename from the original dataset
|
55 |
+
- `name`: Molecular name from QCSchema
|
56 |
+
- `smiles`: Canonical, isomeric SMILES string
|
57 |
+
- `symbols`: Atomic symbols
|
58 |
+
- `atomic_numbers`: Atomic numbers
|
59 |
+
- `geometry`: 3D coordinates in Angstroms
|
60 |
+
- `graph:all`: Molecular connectivity information
|
61 |
+
- `tae@*`: Total Atomization Energy computed with various DFT methods and basis sets
|
62 |
+
- `singlet-triplet-gap-*`: Electronic excitation energies
|
63 |
+
|
64 |
+
## Target Properties
|
65 |
+
|
66 |
+
The dataset includes Total Atomization Energies computed using:
|
67 |
+
|
68 |
+
- **DFT Methods**: B3LYP, B97M-V, M06-2X, r2SCAN, revDSD-PBEP86
|
69 |
+
- **Basis Sets**: def2-SVP, def2-TZVP, def2-QZVP, ma-def2-*
|
70 |
+
- **Composite Methods**: B97-3C, PBEh-3C, r2SCAN-3C
|
71 |
+
- **Semi-empirical**: GFN1-xTB, GFN2-xTB
|
72 |
+
- **High-level**: CCSD(T)/6-31G*, W1-F12
|
73 |
+
|
74 |
+
## Usage Notes
|
75 |
+
|
76 |
+
- **Missing Values**: Many molecules have NaN values for specific methods (sparse coverage)
|
77 |
+
- **SMILES Quality**: Generated from optimized 3D geometries and graph connectivity
|
78 |
+
- **Test Set**: Contains both complete and incomplete records to enable various evaluation strategies
|
79 |
+
|
80 |
+
## Citation
|
81 |
+
|
82 |
+
If you use this processed dataset, please cite both the original MSR-ACC publication and the Zenodo repository:
|
83 |
+
|
84 |
+
```bibtex
|
85 |
+
@misc{ehlert2025accuratechemistrycollectioncoupled,
|
86 |
+
title={Accurate Chemistry Collection: Coupled cluster atomization energies for broad chemical space},
|
87 |
+
author={Sebastian Ehlert and Jan Hermann and Thijs Vogels and Victor Garcia Satorras and Stephanie Lanius and Marwin Segler and Derk P. Kooi and Kenji Takeda and Chin-Wei Huang and Giulia Luise and Rianne van den Berg and Paola Gori-Giorgi and Amir Karton},
|
88 |
+
year={2025},
|
89 |
+
eprint={2506.14492},
|
90 |
+
archivePrefix={arXiv},
|
91 |
+
primaryClass={physics.chem-ph},
|
92 |
+
url={https://arxiv.org/abs/2506.14492},
|
93 |
+
}
|
94 |
+
```
|
95 |
+
|
96 |
+
```bibtex
|
97 |
+
@dataset{ehlert_2025_15783826,
|
98 |
+
author = {Ehlert, Sebastian and
|
99 |
+
Hermann, Jan and
|
100 |
+
Vogels, Thijs and
|
101 |
+
Garcia Satorras, Victor and
|
102 |
+
Lanius, Stephanie and
|
103 |
+
Segler, Marwin and
|
104 |
+
Kooi, Derk P. and
|
105 |
+
Takeda, Kenji and
|
106 |
+
Huang, Chin-Wei and
|
107 |
+
Luise, Giulia and
|
108 |
+
van den Berg, Rianne and
|
109 |
+
Gori-Giorgi, Paola and
|
110 |
+
Karton, Amir},
|
111 |
+
title = {MSR-ACC/TAE25: 77k coupled cluster atomization
|
112 |
+
energies for broad chemical space
|
113 |
+
},
|
114 |
+
month = jul,
|
115 |
+
year = 2025,
|
116 |
+
publisher = {Zenodo},
|
117 |
+
doi = {10.5281/zenodo.15783826},
|
118 |
+
url = {https://doi.org/10.5281/zenodo.15783826},
|
119 |
+
}
|
120 |
+
```
|
121 |
+
|
122 |
+
## License
|
123 |
+
|
124 |
+
This dataset is released under the Community Data License Agreement - Permissive - Version 2.0 (CDLA-Permissive-2.0), consistent with the original MSR-ACC dataset licensing.
|