Predicting unknown gene function is an important problem in biology and bioinformatics. Here we present a machine learning approach using artificial neural networks (ANNs) to predict the small subunit of the terminase protein (TerS) in phages based on length, amino acid distribution and isoelectric point. ANNs were chosen in part because they are able to solve a wide range of difficult classification problems. Data to train the neural networks was taken from the PhAnToMe database of viral and bacterial proteins. The best neural nets generated were then used to make predictions on an independent set of data. The programming language Python was used to sort the training examples in positive and negative sets and the neural nets were generated with the help of MATLAB's neural network toolbox. The results show that neural networks are a promising approach for classifying TerS in a set of genes with unknown function.
