Protein-Protein Interaction Classification Using Jordan Recurrent Neural Network

Proteins form a very important part of a living cell. The biological functions are carried out by the proteins within the cell by interacting with other proteins in other cells. This is called protein-protein interaction. Protein-Protein Interactions are very important in understanding the diseases and finding their cause. It can also provide the basis for new therapeutic approaches. A number of classifiers have been developed till date to classify protein-protein interactions namely SVM, SVM-KNN, Backpropagation Neural Network (BPNN). In this work Jordan Recurrent Neural Network (JRNN) is used to classify the protein-protein interactions. The classifier developed for this work uses amino acid composition of proteins as input to classify the percentage of interacting and non-interacting proteins. The results obtained were best at the threshold value of zero. The classifier gives an accuracy of 97.25% which is 8.7% more than BPNN. The overall accuracy of JRNN for threshold ranging from -1 to +1 with a difference of 0.1 comes out to be 80.1%.