Protein classification using artificial neural networks with different protein encoding methods

The fast growth of annotated biological data implies in the need of developing new techniques and tools to classify these data, in such way that they can be useful. Protein classification is one relevant task in this context. This paper presents different models of neural network, aiming to compare the influence of the protein sequence encoding method in the performance of the Neural network to classify proteins. Besides, it is proposed two methods of protein sequence encoding, that were tested with several neural network, for classifying proteins using two approaches: based on families of proteins and based on function of proteins. The results of performance of the neural networks are presented and compared with other works in the area.