predict DILI scores with machine learning from expression data.
In order to use this code you will have to setup an appropriate python environment. Within this repository there are two anaconda environment files '*.yml', which can be used to easily install all the required packages. If you don't want to use anaconda but rather install the packages through a requirements.txt, see step 1.3.
- Make sure you have Anaconda installed (anaconda.com)
- open Anaconda prompt
- navigate to the directory containing this repository including the *.yml files with the cd command.
- run the following command to install the python environment with all packages:
conda env create -f DILI_prediction_jip.yml
- If you get an error, try the approach in step 1.2 for non-windows 10 platforms (even if you have windows 10)
- If everything went well, run the following command to activate the new python environment:
conda activate keras-cuda2
- Now you're all set, you can type 'spyder' in the anaconda command prompt to start using the code.
- Make sure you have Anaconda installed (anaconda.com)
- open Anaconda prompt
- navigate to the directory containing this repository including the *.yml files with the cd command.
- run the following command to install the python environment with all necessary packages:
conda env create -f DILI_prediction_jip_multi_platform.yml
- If everything went well, run the following command to activate the new python environment:
conda activate keras-cuda2
- Now you're all set, you can type 'spyder' in the anaconda command prompt to start using the code.
- Make sure you have Python 3 installed (this code has been tested only with python 3.6, 3.7 and 3.8)
- navigate to the directory containing this repository including the requirements.txt file with the cd command.
- run the following command:
pip install -r requirements.txt
- Are there any errors? Then the best option would be to install anaconda and use step 1.1/1.2, otherwise install
all packages below manually, and all other packages requested when running the code:
- keras-gpu (or keras if you're not using CUDA)
- tensorflow-gpu (or tensorflow if you're not using CUDA)
- scikit-learn
- pandas
- numpy
- matplotlib (optional)
- seaborn (optional)
The code contains one main file: ‘Assignment_jip.py’. From this file, all the other scripts/functions are called. To get an idea of what this file does, have a look at the 'schematic_overview.svg' file in the repository.
Mass_evaluator.py is not part of the pipeline but was used to quickly run many comparisons and is therefore not properly documented.