AutoSimP: An Approach for Predicting Proteins' Structural Similarities Using an Ensemble of Deep Autoencoders

摘要

This paper investigates the problem of supervisedly classifying proteins according to their structural similarity, based on the information enclosed within their conformational transitions. We are proposing AutoSimP approach consisting of an ensemble of autoencoders for predicting the similarity class of a certain protein, considering the similarity predicted for its conformational transitions. Experiments performed on real protein data reveal the effectiveness of our proposal compared with similar existing approaches.