The 5-fold cross-validation MIPPI models can be used for predicting the impact of a mutation on protein-protein interaction (PPI). You can download the saved weights and do the prediction using the overall results of the five models by adding the prediction scores of each category.

The resource codes can be fetched from the GitHub repository.

The processed IMEx dataset for training and evaluating MIPPI can be downloaded here. You also can dowmload the processed SKEMPI v2 dataset and the processed PsyMuKB dataset for evaluating MIPPI. You need to use the Pandas package (in Python) with 'pandas.read_csv(...)' or 'pandas.read_pickle(...)' to load these datasets.