https://github.com/biomed-AI/CoSMIG

Communicative Subgraph Representation Learning for Multi-Relational Inductive Drug-Gene Interaction Prediction
https://github.com/biomed-AI/CoSMIG

Last synced: 3 months ago
JSON representation

Communicative Subgraph Representation Learning for Multi-Relational Inductive Drug-Gene Interaction Prediction

• Host: GitHub
• URL: https://github.com/biomed-AI/CoSMIG
• Owner: biomed-AI
• License: mit
• Created: 2021-09-27T07:22:31.000Z (about 3 years ago)
• Default Branch: main
• Last Pushed: 2023-02-21T13:07:35.000Z (over 1 year ago)
• Last Synced: 2024-06-29T07:48:03.087Z (4 months ago)
• Language: Python
• Size: 15.9 MB
• Stars: 11
• Watchers: 2
• Forks: 4
• Open Issues: 1
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

• StarryDivineSky - biomed-AI/CoSMIG - 基因相互作用预测的交际子图表示学习 (药物-靶标 药物-药物 化合物-蛋白质 相互作用 / 网络服务_其他)

README

        
# CoSMIG

Communicative Subgraph Representation Learning for Multi-Relational Inductive Drug-Gene Interaction Prediction

![alt text](https://github.com/Jh-SYSU/CoSMIG/blob/main/framework.jpg "Illustration of CoSMIG")

This is the standalone code for our paper: [Communicative Subgraph Representation Learning for Multi-Relational Inductive Drug-Gene Interaction Prediction](https://arxiv.org/abs/2205.05957)

## Requirements

Stable version: Python 3.7.9 + PyTorch 1.7.1+cu110 + PyTorch_Geometric 1.6.3.

Install [PyTorch](https://pytorch.org/)

Install [PyTorch_Geometric](https://rusty1s.github.io/pytorch_geometric/build/html/notes/installation.html)

Other required python libraries: numpy, scipy, pandas, h5py, networkx, tqdm etc.

Also you can  install the required packages follow there instructions (tested on a linux terminal):

`conda env create -f environment.yaml`

## Datasets

Please Contact us ([email protected]) to obtain the Data (from DrugBank and DGIdb) and Splits.

### Statistic of DGI Dataset

|Dataset|DrugBank|DGIdb|

|:-:|:-:|:-:|

|#Drug|425|1185|

|#Gene|11284|1164|

|#Interactions|80924|11266|

|Interaction type|2|14|

## Usages

For training on DrugBank on the transductive scenario:

```

CUDA_VISIBLE_DEVICES=0 python main.py --data-name DrugBank --testing --dynamic-train --dynamic-test --dynamic-val --save-results --max-nodes-per-hop 200

```

For training on DGIdb on the inductive scenario:

```

CUDA_VISIBLE_DEVICES=0 python main.py --data-name DGIdb --testing --mode inductive --dynamic-train --dynamic-test --dynamic-val --save-results --max-nodes-per-hop 200

```

More parameters could be found by:

```

python main.py -h

```

## Reference

If you find the code useful, please cite our paper.

```

@inproceedings{cosmig,

  title     = {Communicative Subgraph Representation Learning for Multi-Relational Inductive Drug-Gene Interaction Prediction},

  author    = {Rao, Jiahua and Zheng, Shuangjia and Mai, Sijie and Yang, Yuedong},

  booktitle = {Proceedings of the Thirty-First International Joint Conference on

               Artificial Intelligence, {IJCAI-22}},

  publisher = {International Joint Conferences on Artificial Intelligence Organization},

  editor    = {Lud De Raedt},

  pages     = {3919--3925},

  year      = {2022},

  month     = {7},

  note      = {Main Track},

  doi       = {10.24963/ijcai.2022/544},

  url       = {https://doi.org/10.24963/ijcai.2022/544},

}

```

## Contact

Jiahua Rao ([email protected]) and Yuedong Yang ([email protected])