Bayesian sequence learning for predicting protein cleavage points

Abstract

A challenging problem in data mining is the application of efficient techniques to automatically annotate the vast databases of biological sequence data. This paper describes one such application in this area, to the prediction of the position of signal peptide cleavage points along protein sequences. It is shown that the method, based on Bayesian statistics, is comparable in terms of accuracy to the existing state-of-the-art neural network techniques while providing explanatory information for its predictions.