Re-training Machine Learning Models using Jupyter Notebooks
A Jupyter Notebook is provided to generate/retrain classification models for six transporter proteins (BCRP, BSEP, OATP1B1, OATP1B3, MRP3, P-gp). Four different classifiers can be selected without an extensive descriptor selection and hyperparameter search as they have been pre-selected. An in-house data set can be combined with the available UNIVIE data set(s) to extend the chemical space of the model.
Getting Started
- Download files from the repository
- Install Anaconda on your device
- Install/activate the virtual environment
- Start the Jupyter Notebook
There are four sections available in the code:
- Data Collection
- Data Set Preparation for ML Task
- Applicability Domain
- Model Generation & Evaluation
Please follow the instructions in file Tutorial_Jupyter_Notebook.pdf. A star next to the title/header indicates that an action from your side is required.