https://github.com/chao1224/biochemgnn_dense
https://github.com/chao1224/biochemgnn_dense
Last synced: 3 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/chao1224/biochemgnn_dense
- Owner: chao1224
- Created: 2020-06-26T00:51:36.000Z (over 5 years ago)
- Default Branch: master
- Last Pushed: 2021-01-28T23:53:51.000Z (almost 5 years ago)
- Last Synced: 2025-04-02T16:25:01.321Z (7 months ago)
- Language: Python
- Size: 206 KB
- Stars: 11
- Watchers: 5
- Forks: 1
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# BioChem Graph Neural Network --- Dense
This is the dense version of the Graph Neural Network for biology and chemistry molecules. The `dense` means each graph is padded to length of N, where N is the largest size of molecules in this dataset.
## Environment
```
conda create -n benchmark python=3.7
source activate benchmark
conda install -y -c pytorch pytorch=1.5 torchvision
conda install -y -c rdkit rdkit
conda install -y scikit-learn
conda install -y numpy
conda install -y matplotlib
```
## Models
| Model | Paper |
| :---: | :---: |
| ECFP | The generation of a unique machine description for chemical structures-a technique developed at chemical abstracts service, ACS Journal of Chemical Documentation 1965 |
| Neural Fingerprint (NEF) | Molecular graph convolutions: moving beyond fingerprints, NeurIPS 2016 |
| Graph Convolutional Network (GCN) | Semi-Supervised Classification with Graph Convolutional Networks, ICLR 2017 |
| Deep Tensor Neural Networks (DTNN) | Quantum-chemical insights from deep tensor neural networks, Nature Communications 2017 |
| Edge Neural Network (ENN) / MPNN | Neural Message Passing for Quantum Chemistry, ICML 2017 |
| SchNet | SchNet: A continuous-filter convolutional neural network for modeling quantum interactions, NeurIPS 2017 |
| Graph Isomorphism Network (GIN) | How Powerful are Graph Neural Networks?, ICLR 2019 |
| Directed Message Passing Neural Network (D-MPNN) | Analyzing Learned Molecular Representations for Property Prediction, ACS JCIM 2019 |
## Supervised Experiments
### random split
| | ECFP | NEF | DTNN | ENN | GCN | GIN | SchNet | D-MPNN |
| :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: | :---: |
| BACE | Done | Done | Done | Done | Done | Done | Done | Done |
| BBBP | Done | Done | Done | Done | Done | Done | Done | Done |
| Delaney | Done | Done | Done | Done | Done | Done | Done | Done |
| FreeSol | Done | Done | Done | Done | Done | Done | Done | Done |