Estimate patient's outcome from cancer RNAseq using deep learning models trained with transfer learning.
These instructions will get you a copy of the project up and running on your local machine.
- Docker deamon up and running
- Docker Compose
The file DeepIO/.env holds all of the configurations of the server. Edit it to adjust the different settings. The DOMAIN variable is the address of the machine that will host the server. The JWT variable is the secret token.
Add your models (.h5 et .RData) to DeepIO/backend/prediction_deamon/ML_models/.
To launch the server, simply type:
./start_server.sh
If the Docker deamon was started with sudo, run:
sudo ./start_server.sh
To use the application, open your web browser and go to the following adress (replace the content between brackets by the values in DeepIO/.env):
http://[DOMAIN]:[FRONTEND_PORT]
All the data gathered and created by the application are stored in the directory DeepIO/app_memory at the root of the application. The uploaded files and the MongoDB database will be stored in it. It means that you can use this folder to export your data or use data from another instance of the application.
Most recent versions of Tensorflow use CPU instructions that some old CPUs does not have. This can result in error when starting the server (in the backend part). To solve this issue, edit the DeepIO/backend/venv_requirement.txt file by replacing the lines:
Keras==2.3.1
tensorflow==1.14.0
by:
Keras==2.2.0
tensorflow==1.5.0