An intelligent computer-aided approach for target protein prediction in infectious diseases

Essential proteins are the most important constituents for all kinds of organism. The research on predicting essential target proteins contributes significantly to drug development, disease diagnosis, and treatment. Numerous computational- and experimental-based approaches have been used in the recent past for essential protein prediction. However, it is highly challenging to bring remarkable improvement in the accuracy of the essential target proteins. In this method, we introduce an intelligent computational technique known as graph colouring-deep neural network which automatically extracts the target proteins by combining a graph-theoretic approach called graph colouring and neural network. In this approach, the protein–protein interaction network (PPI) of homosapiens are extracted from string DB (data base) and then applied to the graph colouring algorithm. Initially, each protein in the network is assigned a colour by checking the connectivity with the neighbourhood proteins. Secondly, the proteins with primary and secondary colours are extracted from the PPI. Finally, a deep neural network-based approach is used to automatically extract the essential target proteins depending on the physicochemical features of the proteins. To assess the performance of the proposed model, the experiment has been carried out in four different diseases such as cancer, diabetes, asthma and human papilloma virus viral infection. The proposed approach shows a remarkable performance than the traditional approaches in views of various metrics such as accuracy, precision, recall, and F-measure.