PROTEIN-PROTEIN INTERACTION PREDICTION USING A DEEP NEURAL NETWORK WITH BATCH NORMALIZATION AND QUARTILE ALGORITHM

Detecting protein-protein interactions (PPIs) is key for disease therapy development. While experimental methods are costly, deep neural network (DNN) models now use available PPI data for prediction, though limited by low-quality sequence-based data. This study introduces FDPPI, a DNN model leveraging a quartile-based algorithm and batch normalization to enhance performance, achieving 98.09% accuracy, 98.34% precision, and 97.72% sensitivity on human PPI data.