ResNetIC50

A ResNet model for drug response prediction, using cell viability inhibitory concentrations (IC50 values)
ResNetIC50 is the ResNet model constructed from scenario 1 (MDG-160K).

Usage

Jupyter notebook files of python coding for each scenarios are stored in Model_generation_validation
All the dataset and pre-defined models of each scenarios can be downloaded from here (IC50evaluation/Dataset, IC50evaluation/Pre-defined_models)
You should download each dataset and change the path in each code (path for dataset, model_output_folder, result_output_folder, etc)
Please set and check the file path (workdir) where dataset and training-test set split file located and for model and output file in each code.
Please refer to the annotations in each code.

*.ipynb : the jupyter notebook files (python 3) to construct prediction model and show validation result
*.h5, *.json : these files are constructed model architecture and model weight from AI-method (CNN and ResNet)
*.pkl : these files are constructed model from ML-method (linearSVR, random forest, XGBoost, etc)