Predicting drug synergy in breast cancer with deep learning using target-protein inhibition profiles

In this study, a 3×3 nested cross-validation (CV) method was used. The initial set of drug combinations, which consisted of 24145 pairs, were divided into three folds in the outer loop of the nested CV, with one fold serving as the test dataset while the other two folds were further divided into three folds in the inner loop. During each round of the inner loop, two folds were used as a training dataset, while the other fold was used as a validation set in a grid search for the best hyperparameter set. The datasets and Python codes required for this task can be found in the 'identify_best_parameters' folder. To perform a grid search on each round of the inner loops, execute GridSearch.py in each subfolder.

The best hyperparameter set (identified based on the average Pearson correlation coefficients obtained across each round of the three inner loops) was used to train a model (see Table 4 in the main text), and the model was evaluated using the test set from the outer loop. The datasets and Python codes required to train the three final models are provided in the 'train_final_models' folder.