Accurate classification of membrane protein types based on sequence and evolutionary information using deep learning

Lei Guo,Mingyuan Li,Shunfang Wang,Zicheng Cao

doi:10.1186/s12859-019-3275-6

Lei Guo, Mingyuan Li + Show 2 more

Open Access

https://doi.org/10.1186/s12859-019-3275-6

Copy DOI

Journal: BMC Bioinformatics	Publication Date: Dec 1, 2019
Citations: 28	License type: open-access

Affiliation: Yunnan University, Sun Yat-sen University

Abstract

BackgroundMembrane proteins play an important role in the life activities of organisms. Knowing membrane protein types provides clues for understanding the structure and function of proteins. Though various computational methods for predicting membrane protein types have been developed, the results still do not meet the expectations of researchers.ResultsWe propose two deep learning models to process sequence information and evolutionary information, respectively. Both models obtained better results than traditional machine learning models. Furthermore, to improve the performance of the sequence information model, we also provide a new vector representation method to replace the one-hot encoding, whose overall success rate improved by 3.81% and 6.55% on two datasets. Finally, a more effective model is obtained by fusing the above two models, whose overall success rate reached 95.68% and 92.98% on two datasets.ConclusionThe final experimental results show that our method is more effective than existing methods for predicting membrane protein types, which can help laboratory researchers to identify the type of novel membrane proteins.

Highlights

Membrane proteins play an important role in the life activities of organisms
The performance of vector representation To explore the advantage of the new vector representation method, we compared the performance between the onehot encoding and our method in the sequence information model
To verify whether the vector representation that we proposed is better than one-hot encoding, we indicated the overall success rate of the one-hot encoding method in the figures

Summary

Introduction

Membrane proteins play an important role in the life activities of organisms. Knowing membrane protein types provides clues for understanding the structure and function of proteins. Guo et al BMC Bioinformatics 2019, 20(Suppl 25):700 According to their functions, membrane proteins can be classified into three classes: integral, peripheral and lipidanchored [9]. To help laboratory researchers discover the type of novel membrane protein, various computation methods are proposed for membrane protein type recognition. Many of these approaches incorporate machine learning algorithms and statistical analysis techniques, such as knearest neighbor (KNN) [10], the naive Bayesian model (NBM) [11], support vector machines (SVM) [12,13,14], random forests (RF) [15], probabilistic neural network (PNN) [16] and hidden Markov models [7]

Methods

Results

Conclusion