Metadata-Version: 2.1
Name: grape-chem
Version: 1.0.4
Summary: Tools for computational chemistry and deep learning.
Author-email: Felix Aertebjerg <aertebjerg.felix@gmail.com>
Project-URL: Homepage, https://github.com/aerte/GraPE
Classifier: Programming Language :: Python :: 3
Classifier: License :: OSI Approved :: MIT License
Classifier: Operating System :: OS Independent
Requires-Python: >=3.9
Description-Content-Type: text/markdown
License-File: LICENSE
Requires-Dist: torch==2.1.0
Requires-Dist: torch_geometric>=2.1
Requires-Dist: torch-scatter
Requires-Dist: rdkit~=2023.9.5
Requires-Dist: dgl~=2.0.0
Requires-Dist: pandas~=2.2.0
Requires-Dist: numpy~=1.26.3
Requires-Dist: pyarrow
Requires-Dist: openpyxl
Requires-Dist: dgllife~=0.3.2
Requires-Dist: matplotlib~=3.8.3
Requires-Dist: seaborn~=0.13.2
Requires-Dist: CIRpy~=1.0.2
Requires-Dist: tqdm~=4.66.1
Requires-Dist: scikit-learn

GraPE-Chem - Graph-based Property Estimation for Chemistry
===========================================================

This is a python package to support Chemical property prediction using [PyTorch](https://pytorch.org/docs/stable/index)
and [PyTorch Geometric](https://pytorch-geometric.readthedocs.io/en/latest/).
The ambition of this project is to build a flexible pipeline that lets users go from molecule
descriptors (SMILES) to a fully functioning Graph Neural Network and allow for useful customization
at every step.

For more information, please check out the [docs](https://grape-chem.readthedocs.io/en/latest/).



Installing the toolbox
----------------------

To use the package, please run the following inside a terminal:

``pip install grape-chem``


Demonstrations and Use
-----------------------
After installing, the package will work like any other. See ``Demo``
and ``Advanced Demo`` inside of [docs](https://grape-chem.readthedocs.io/en/latest/) 
for an introduction of how the toolbox can be used.



Note
-----
If optimization is run on hpc using `GraPE` and the optimization procedure outlined in
the ``Advanced Demonstration``, the following requirements need to be met:

``
python==3.9
``
``
cuda==12.1
``

and the following package need to be re-installed using the correct cuda-version:

``
torch==2.1.2
``
``
dgl~=1.1.3
``
``
torch-scatter -f https://data.pyg.org/whl/torch-2.1.2+cu121.html
``
``
ray
``
``
ConfigSpace==0.4.18
``
``
hpbandster==0.7.4
``

The reason for the particular python version is a subpackage in ``hpbandster``.



