GRM:Gaussian-enhanced Representation Model for Extracting Protein-Protein Interactions Affected by Mutations

The implementation of method in our paper: Extracting Protein-Protein Interactions Affected by Mutations via Gaussian-enhanced Representation and Contrastive Learning

1.Prepareration

requirements

requirements are listed in requirements.txt

DataSet & Evaluation

Original dataset and evaluation scripts can be downloaded here,

And some of annotation of genes are modified as shown in the end of the document

The modified datasets and the results obtained by GNP are available here.

2.Reproducing Results

10-fold cross validation on train

RC-Only: bash scripts/train_RC.sh

Train and eval on test

RC-Only: bash scripts/test_RC.sh

The result of RC with confidence will be saved as ./outputjson/{loggercomment}_{epoch}.json.

Analysis

1. analysis.ipynb

   Predicted relations need to be post-processed here before homolo eval and exact eval.

   Scripts about case study can be found here.

2. python cross_fold_metrics.py > metrics.tsv print results of the cross fold validation to metrics.tsv