|
|
|
|
|
|
|
|
|
|
|
## Train Mogonet |
|
|
|
|
|
|
|
|
|
|
|
Having created the input matrices, we should feed the data to Mogonet architecture. Following these steps: |
|
|
|
|
|
|
|
|
|
|
|
1. Clone the repository: (preferably in Colab or a server with GPU available) |
|
|
|
|
|
|
|
|
|
|
|
```bash |
|
|
|
|
|
git clone https://github.com/txWang/MOGONET.git |
|
|
|
|
|
``` |
|
|
|
|
|
|
|
|
|
|
|
2. Configure the code. To do so, copy the output from previous stage to the `MOGONET` directory and add the directory name and info to `main_biomarker.py` and `main_mogonet.py`. |
|
|
|
|
|
|
|
|
|
|
|
3. Run `main_mogonet.py` to train the network and store the model |
|
|
|
|
|
|
|
|
|
|
|
4. Run `main_biomarker.py` to evaluate the feature importances and extract biomarkers from the model. |