Artificial neural network to prediction of protein-protein interactions in yeast

The problems of predicting the Protein-Protein Interactions (PPIs) are characterized by probabilistic constraints using the artificial neural network techniques. In the literature, no specific rules are proposed for determining whether two proteins interact, but various approaches have been proposed to collect the information about the interaction between the proteins. The need and importance of PPIs, their interactions, accurate computational PPI determination and prediction of PPIs based on amino acid coding of its physiochemical properties motivate the researchers to propose a new methodology. This paper presents the simulation steps for designing the prediction methodology of PPI in the framework of artificial neural network (ANN) learning optimization. Our methodology includes six simulation steps for determining the interaction, where the publically available data set is compared with the Gold dataset to identify the interactions. The results of PPI simulation with generated feature vector and learning routines of ANN suggests that the best parameters are obtained through our proposed methodology produce better accuracies compared to others.