Email updates

Keep up to date with the latest news and content from BMC Bioinformatics and BioMed Central.

Open Access Highly Accessed Methodology article

Quantum coupled mutation finder: predicting functionally or structurally important sites in proteins using quantum Jensen-Shannon divergence and CUDA programming

Mehmet Gültas12*, Güncel Düzgün1, Sebastian Herzog1, Sven Joachim Jäger1, Cornelia Meckbach1, Edgar Wingender2 and Stephan Waack1*

Author Affiliations

1 Institute of Computer Science, University of Göttingen, Goldschmidtstr. 7, 37077 Göttingen, Germany

2 Institute of Bioinformatics, University of Göttingen, Goldschmidtstr. 1, 37077 Göttingen, Germany

For all author emails, please log on.

BMC Bioinformatics 2014, 15:96  doi:10.1186/1471-2105-15-96

Published: 3 April 2014

Abstract

Background

The identification of functionally or structurally important non-conserved residue sites in protein MSAs is an important challenge for understanding the structural basis and molecular mechanism of protein functions. Despite the rich literature on compensatory mutations as well as sequence conservation analysis for the detection of those important residues, previous methods often rely on classical information-theoretic measures. However, these measures usually do not take into account dis/similarities of amino acids which are likely to be crucial for those residues. In this study, we present a new method, the Quantum Coupled Mutation Finder (QCMF) that incorporates significant dis/similar amino acid pair signals in the prediction of functionally or structurally important sites.

Results

The result of this study is twofold. First, using the essential sites of two human proteins, namely epidermal growth factor receptor (EGFR) and glucokinase (GCK), we tested the QCMF-method. The QCMF includes two metrics based on quantum Jensen-Shannon divergence to measure both sequence conservation and compensatory mutations. We found that the QCMF reaches an improved performance in identifying essential sites from MSAs of both proteins with a significantly higher Matthews correlation coefficient (MCC) value in comparison to previous methods. Second, using a data set of 153 proteins, we made a pairwise comparison between QCMF and three conventional methods. This comparison study strongly suggests that QCMF complements the conventional methods for the identification of correlated mutations in MSAs.

Conclusions

QCMF utilizes the notion of entanglement, which is a major resource of quantum information, to model significant dissimilar and similar amino acid pair signals in the detection of functionally or structurally important sites. Our results suggest that on the one hand QCMF significantly outperforms the previous method, which mainly focuses on dissimilar amino acid signals, to detect essential sites in proteins. On the other hand, it is complementary to the existing methods for the identification of correlated mutations. The method of QCMF is computationally intensive. To ensure a feasible computation time of the QCMF’s algorithm, we leveraged Compute Unified Device Architecture (CUDA).

The QCMF server is freely accessible at http://qcmf.informatik.uni-goettingen.de/ webcite.