Volume 12 Supplement 1
HDX-Analyzer: a novel package for statistical analysis of protein structure dynamics
- Sanmin Liu†1, 2, 3,
- Lantao Liu†1, 3, 4,
- Ugur Uzuner†1, 4,
- Xin Zhou†1,
- Manxi Gu5,
- Weibing Shi1, 4,
- Yixiang Zhang1, 4,
- Susie Y Dai2, 6Email author and
- Joshua S Yuan1, 4Email author
© Liu et al; licensee BioMed Central Ltd. 2011
Published: 15 February 2011
HDX mass spectrometry is a powerful platform to probe protein structure dynamics during ligand binding, protein folding, enzyme catalysis, and such. HDX mass spectrometry analysis derives the protein structure dynamics based on the mass increase of a protein of which the backbone protons exchanged with solvent deuterium. Coupled with enzyme digestion and MS/MS analysis, HDX mass spectrometry can be used to study the regional dynamics of protein based on the m/z value or percentage of deuterium incorporation for the digested peptides in the HDX experiments. Various software packages have been developed to analyze HDX mass spectrometry data. Despite the progresses, proper and explicit statistical treatment is still lacking in most of the current HDX mass spectrometry software. In order to address this issue, we have developed the HDXanalyzer for the statistical analysis of HDX mass spectrometry data using R, Python, and RPY2.
Implementation and results
HDXanalyzer package contains three major modules, the data processing module, the statistical analysis module, and the user interface. RPY2 is employed to enable the connection of these three components, where the data processing module is implemented using Python and the statistical analysis module is implemented with R. RPY2 creates a low-level interface for R and allows the effective integration of statistical module for data processing. The data processing module generates the centroid for the peptides in form of m/z value, and the differences of centroids between the peptides derived from apo and ligand-bound protein allow us to evaluate whether the regions have significant changes in structure dynamics or not. Another option of the software is to calculate the deuterium incorporation rate for the comparison. The two types of statistical analyses are Paired Student’s t-test and the linear combination of the intercept for multiple regression and ANCOVA model. The user interface is implemented with wxpython to facilitate the data visualization in graphs and the statistical analysis output presentation. In order to evaluate the software, a previously published xylanase HDX mass spectrometry analysis dataset is processed and presented. The results from the different statistical analysis methods are compared and shown to be similar. The statistical analysis results are overlaid with the three dimensional structure of the protein to highlight the regional structure dynamics changes in the xylanase enzyme.
Statistical analysis provides crucial evaluation of whether a protein region is significantly protected or unprotected during the HDX mass spectrometry studies. Although there are several other available software programs to process HDX experimental data, HDXanalyzer is the first software program to offer multiple statistical methods to evaluate the changes in protein structure dynamics based on HDX mass spectrometry analysis. Moreover, the statistical analysis can be carried out for both m/z value and deuterium incorporation rate. In addition, the software package can be used for the data generated from a wide range of mass spectrometry instruments.
Protein intrinsic dynamics has been more and more recognized as an important consideration for protein functions . Several recent studies have revealed that protein dynamics plays essential roles in the catalysis and other functions [1, 2]. Among the different techniques, HDX mass spectrometry stands out as a relatively high throughput platform to probe the backbone dynamics of the proteins [3–6]. HDX mass spectrometry has been broadly applied to study protein dynamics and structure, particularly for the protein binding with ligands, substrates, DNA and other molecules [4, 7–10]. Such analysis has enabled the illustration of mechanisms for enzyme substrate interaction and the molecular determinants during protein binding [11, 12]. The fundamental concept of HDX mass spectrometry analysis is based on the mass increase of a protein when the protein protons exchanged with solvent deuterium . The rate and percentage of the H/D exchange can be measured by mass to charge ratio (m/z) of the protein. The HDX mass spectrometry can be utilized to study both the global and regional protein conformational changes [13, 14]. Coupled with protein digestion and chromatography separation, the HDX mass spectrometry enables characterizing different regions of protein for H/D exchange based on the peptide H/D exchange rate or the m/z of the peptides. In a differential HDX experiment, typically two protein forms will be compared. The apo protein and the ligand bound protein are subjected to HDX experiment in a parallel mode. The information allows one to understand which region of the protein is more stabilized or destabilized upon ligand binding in the solvent exchange reaction [3–5, 14–22]. If more H/D exchange is observed in a particular region, the protein region is more dynamic in the solvent exchange reaction, meaning that the region is more flexible or less stable in HDX. In a typical differential HDX experiments, the protein of interest is subject to different exchange times in its apo form and protein ligand complex with technical replicates. The data processing for HDX mass spectrometry thus requires us to compare a large set of the m/z values or percentages of deuterium incorporation for the same peptides derived from apo protein and ligand bound protein.
Various software platforms have been developed to analyze the HDX data. Among them include HX-Express, Deuterator, HD Desktop, DEX, Hydra, TOF2H etc. Most of these HDX data analysis software packages focus on calculating the m/z value from the MS raw data for the deuterated peptide, and then evaluate the m/z value increase according to time. For example, HX-Express is a semi-automated software package which exports deuterium uptake curve and peak width plots based on Microsoft excel application . Compared with HX-Express, Deuterator is more automated and can deconvolute overlapping mass peaks. Meanwhile, Deuterator acts as a web-based server to process HDX data sets on line . Furthermore, HD Desktop is built on top of deuterator, and integrates more tools for data extraction displaying visualization components . DEX uses a Fourier deconvolution method for computing high-resolution mass spectrometry data . Hydra executes through a user-defined workflow, by which deuterium incorporation values are extracted and visualized in tabular and graphical formats. Hydra also automates the extraction and visualization of deuterium distribution values for large data sets . TOF2H focuses on interpreting MALDI-based HDX data and also builds up a pipeline for automated data processing . Despite the significant progresses mentioned above, most software uses absolute differences between HD exchange rates as an evaluation of the differential structure dynamics changes, whereas few tools enables statistical evaluation of differential HD exchange among different conditions. It is noted that CalcDeut  evaluate the statistical distribution of deuterium incorporated into protease digested peptide fragments to compensate data truncation due to instrument signal to noise ratio. Hydra provides a prototype of student t-test evaluation and calculates a p-value for the differential HD exchange. Despite the significant advances made by Hydra, due to the inherent limitation of regular Student’s t-test, the statistical treatment provided by Hydra is suitable to analyze data at one exchange time point, though HDX experiments always involve multiple time points and the HDX rate is time-dependent. In order to address the issue, we integrated the pairwise t-test, multiple regression and ANCOVA models to provide the accurate evaluation of the differential HD exchange for proteins under different treatments. It needs to be realized that the HDX exchange of a peptide is time-dependent, and the usage of regular Student’s t-test to compare the same peptide throughout the whole HDX time course is not accurate. A pairwise t-test can be used. A multiple regression model or ANCOVA model is more suited for this scenario. The statistical analysis of differential exchange rate for peptide is crucial for evaluating if significant differential structural dynamics changes exist for a specific peptide region or not. In many studies, a large difference in exchange rate may not reflect the differential structure dynamics changes if the standard deviation for the exchange rate is high. For this reason, we have developed a new platform offering various choices for statistical evaluation of HD exchange rate among different samples.
The integration of statistical analysis with data processing is challenging. In terms of statistical analysis, several software environment including SAS, SPSS, and R can be employed. Among these packages, R is the open source program and can be easily obtained from internet free of charge. Despite the various advantages of R, the software environment does not have strong user-interface supports and thus requires certain level of expertise. In order to develop user-friendly statistical software for HDX mass spectrometry analysis, we hereby employ the latest RPY2 package to connect the statistical module of R with a data processing module implemented by Python and a user-interface implemented by wxpython. Many programming languages including C, java, Perl, Python can be used for data processing and UI development, and each programming language has its pros and cons. Among these programming languages, Python is chosen as the main developing language for two reasons: First, the existed RPY package allowed the seamless and effective implementation between the data processing module and statistical module in R. Second, various BioPython packages have been developed for the analysis of biological data which allows fast and easy embedding. For these reasons, we have developed the HDX mass spectrometry analysis software HDXanalyzer using Python, R, and RPY2 packages.
In this article, we hereby present a novel software package HDXanalyzer for statistical analysis of HDX mass spectrometry data to evaluate the protein structure dynamics changes. The software package includes three major components, the data feeding and processing module, the user interface, and the statistical analysis module. The data processing module is developed in Python to process a batch of excel input files containing the m/z value of the peptides from different experiments. The pre-formatted m/z value of the peptides will be processed to derive the centroid of the peptide peaks or the percentage of deuterium incorporation for the statistical analysis. The data is then processed by the Figure Generator to create graphs to visualize the differential HD exchange rates in apo and ligand bound protein. Further statistical analysis is carried out by R, where two statistical methods are used. The Paired Student’s t-test is used to compare either the centroid values of the m/z value or the deuterium incorporation rate to derive point estimation, confidence intervals, and p value to indicate if significant differences in structure dynamics exist or not. In addition, the multiple regression (or ANCOVA) model is also involved for the similar analysis through linear combination of the intercepts. HDXanalyzer thus provided novel solutions toward ultimate quantification and statistical evaluation of the structure dynamics changes in the HDX mass spectrometry experiments. The software package addressed the imminent need of statistical evaluation for the HDX mass spectrometry analysis and can be expanded to other applications for HD exchange studies by other techniques.
I. Data processing as implemented by python
where m/z (P), m/z (N), and m/z (F) are the centroid values of partially deuterated peptide, nondeuterated peptide, and fully deuterated peptide, respectively .
The resulted data are then processed into a table format and loaded to Figure Generator to create visualization of the dynamic status of peptide at different time points. Specifically, the figures are graphs displaying the m/z value or deuterium incorporation rate of a peptide at different exchange time. Gnuplot, an open source GNU plotting tool under UNIX/linux, with counterpart in MSDOS & Windows system, was employed to implement the Figure Generator. The advantages of Gnuplot lies in two aspects including the availability from either GNU projects or internet free of charge, as well as the convenience of automated generating multiple outputs using its corresponding scripting language. Besides the graphic display of the HDX data, statistical analysis is carried out to generate the point estimation for differential m/z value or incorporation rate, the confidence intervals and the p value.
II. Statistical models and implementation
Statistical analysis is employed to evaluate if a peptide or a region of the protein has significant changes in structure dynamics or not. Such changes are reflected in the differences of either centroid m/z values or the deuterium incorporation rates during the HDX experiments. The m/z value or deuterium incorporation rate from different peptides can be compared with different statistical models to derive parameter estimation and p value. The parameter estimations allow us to evaluate the levels and variations of the differences in structure dynamics of a protein region, and the p value allows us to determine if the differences are significant or not.
Two types of statistical models are used. First, a Paired Student’s t-test (Pairwise t-test) is used to compare the m/z value or percentage of deuterium incorporation for peptides from apo or ligand binding proteins. Paired Student’s T-test is used instead of the regular T-test because of the time effects in the HDX experiments. More specifically, the m/z values or the percentages of deuterium incorporation for a peptide will increase as the hydrogen deuterium exchange time gets longer. As these two values are time dependent, both will reach a plateau when the exchange time is long enough. For this reason, the Paired t-test is used to avoid the time point effects. Besides the pairwise t-test, the multiple regression model or ANCOVA model are utilized to compare the m/z or incorporation rate differences between peptides from the two types of proteins (apo and ligand bound). The multiple regression model is as shown in Equation 2. The linear combination of the Group effects allows us to compare the differences between apo and ligand-bound proteins. For either model, the point estimation of mean differences, confidence intervals, and p value will be computed.
Y = β T X Time + β G X Group + β TG X Time *X Group Equation 2
where Y is the dependent variable that can be either the m/z value or the deuterium incorporation rates of different peptides. Y is dependent on the effects of time points and different groups from either apo or ligand bound proteins. The combination of the two effects may also influence the dependent variable.
III. RPY for integrating the different components
The integration of statistical analysis, data processing, and visualization is usually challenging. The recent developed RPY allows us to integrate the statistical feature of R and the user interface as well as the data processing features of Python. As an open-source language, R has the unique advantages over other statistical languages for software development. RPY enables us to employ the R for statistical analysis of HDX mass spectrometry data. We have also used RPY2 to provide a low-level interface to R. The Python-based system thus can directly call R function through RPY and the software efficiency and effectiveness are greatly improved.
IV. User interface as implemented by wxpython
Result and discussion
HDXanalyzer is implemented as a software package to enable the statistical analysis of HDX mass spectrometry data and to allow the evaluation of protein structure dynamics changes. In order to demonstrate the software application, we first analyze a previously published dataset for the HDX mass spectrometry analysis of xylanase enzyme. The example data is available in the online supplementary document. Furthermore, we also apply HDXanalyzer to analyze two peptides from a recent publication, where the HD exchange for the two peptides were statistically evaluated. We hereby discuss the usage of the software, present the output, compare the different results from different statistical models, and interpret the results.
I. The input data format and the usage of the package
The HDXanalyzer aims to integrate statistical analysis for comparing structure dynamics of protein upon ligand or substrate binding. As discussed in the Implementation section, the software takes a batch of pre-formatted excel files containing m/z values for multiple peptides of different treatment and time points as shown in Supplementary File 1 available online (the HDX_Xylohexaose.rar dataset). The data pre-formatting will allow the software to process a uniform input of HDX mass spec data from different instruments. The sample input file is derived from a xylanase structure dynamics study and the m/z values of the peak area for the peptides are included. Each input excel file will contain several sheets for the data from different peptides and treatments. The spreadsheet contains peptide ID, m/z value, charge state, time points for deuterium treatment, and the ligand name to separate different experimental sets, e.g., apo set and ligand set. The peptide ID can be corresponding to a certain peptide sequence. The upload function is available from the user interface, where input file can be read and processed to generate m/z centroids and deuterium incorporation rates of the peptides as aforementioned. The data are therefore further analyzed for visualization and statistical analysis.
II. The output of HDXanalyzer
III. The interpretation and comparison of different statistical models
As aforementioned, two types of statistical analyses are implemented for HDXanalyzer. Both Paired Student’s T-test and the linear combination of intercept for group (apo vs. ligand) effects in multiple regression (ANCOVA) model are used to derive parameter and p value estimation. A very important decision for data processing is the choice of time point. The early time points after deuterium exchange, especially for exchange less than 1 minute, may lead to very limited exchange even in apo protein. In such case, the statistical analysis of data from these time points cannot represent the real deuterium incorporation level. Therefore, HDXanalyzer offers the users the option to choose the time points for the analysis and we have focused on the time points after ten minutes in our analysis of the example dataset.
The comparison of statistical analysis results from Paired Student’s t-test and multiple regression.
IV. The adaptability to data from different instrument sources
Another strength of HDXanalyzer is that the data pre-processing allows the package to analyze HDX mass spectrometry data from a wider range of instruments. Most of the current software packages are developed on one type of instrument or another and are more adapted toward the high resolution data from very high-end instruments. In order to allow HDXanalyzer to process data from various instruments, we decide to handle the pre-processed data as shown in the supplementary file. The data pre-processing step will be able to handle the HDX mass spectrometry data from different instrument types and the pre-processed data can then be analyzed by HDXanalyzer regardless of instrument types.
V. The overlay of 3D structure for differential structure dynamics of xylanase
HDXanalyzer as a statistical analysis software has enabled the accurate evaluation of the changes of protein structure dynamics. The software integrates the graphic visualization and statistical analysis to enable the effective evaluation of the differential structure dynamics in the HDX mass spectrometry experiments.
The research is supported by the Southcentral Sungrant awarded to SYD and JSY, the startup fund and bioenergy initiatives for JSY and SYD from Texas Agrilife.
This article has been published as part of BMC Bioinformatics Volume 12 Supplement 1, 2011: Selected articles from the Ninth Asia Pacific Bioinformatics Conference (APBC 2011). The full contents of the supplement are available online at http://www.biomedcentral.com/1471-2105/12?issue=S1.
- Henzler-Wildman K, Kern D: Dynamic personalities of proteins. Nature 2007, 450(7172):964–972. 10.1038/nature06522View ArticlePubMedGoogle Scholar
- Uzunner U, Shi WB, Liu L, Liu S, Dai SY, Yuan JS: Enzyme structure dynamics of xylanase I from Trichoderma longibrachiatum . BMC Bioinformatics 2010, 11(Suppl 6):S12. 10.1186/1471-2105-11-S6-S12View ArticleGoogle Scholar
- Dai SY, Chalmers MJ, Bruning J, Bramlett KS, Osborne HE, Montrose-Rafizadeh C, Barr RJ, Wang Y, Wang MM, Burris TP, et al.: Prediction of the tissue-specificity of selective estrogen receptor modulators by using a single biochemical method. Proc Natl Acad Sci U S A 2008, 105(20):7171–7176. 10.1073/pnas.0710802105PubMed CentralView ArticlePubMedGoogle Scholar
- Englander SW: Hydrogen exchange and mass spectrometry: A historical perspective. J Am Soc Mass Spectrom 2006.Google Scholar
- Krishna MM, Hoang L, Lin Y, Englander SW: Hydrogen exchange methods to study protein folding. Methods 2004, 34: 51–64. 10.1016/j.ymeth.2004.03.005View ArticlePubMedGoogle Scholar
- Roder H, Elove GA, Englander SW: Structural characterization of folding intermediates in cytochrome c by H-exchange labelling and proton NMR. Nature 1988, 335(6192):700–704. 10.1038/335700a0PubMed CentralView ArticlePubMedGoogle Scholar
- Li J, Lim MS, Li S, Brock M, Pique ME, Woods VL, Craig L: Vibrid cholerae toxin-coregulated pilus structure analyzed by hydrogen/deuterium exchange mass spectrometry. Structure 2008, 16(1):137–148. 10.1016/j.str.2007.10.027PubMed CentralView ArticlePubMedGoogle Scholar
- Brock M, Fan F, Mei FC, Li S, Gessner C, Woods VL, Cheng X: Conformational analysis of Epac activation using amide hydrogen/deuterium exchange mass spectrometry. J Biol Chem 2007, 282(44):32256–32263. 10.1074/jbc.M706231200View ArticlePubMedGoogle Scholar
- Lambris JD, Sfyroera G, Schuster M, Chen H, Tzekou A, Papp K, Winters M, Woods VL: Studies on the solvent accessibility of native C3 and its fragments, as analyzed by HDX-MS. Mol Immunol 2007, 44(1–3):202–202. 10.1016/j.molimm.2006.07.130View ArticleGoogle Scholar
- Eyles SJ, Kaltashov IA: Methods to study protein dynamics and folding by mass spectrometry. Methods 2004, 34(1):88–99. 10.1016/j.ymeth.2004.03.015View ArticlePubMedGoogle Scholar
- Begley MJ, Taylor GS, Brock MA, Ghosh P, Woods VL, Dixon JE: Molecular basis for substrate recognition by MTMR2, a myotubularin family phosphoinositide phosphatase. Proc Natl Acad Sci U S A 2006, 103(4):927–932. 10.1073/pnas.0510006103PubMed CentralView ArticlePubMedGoogle Scholar
- Derunes C, Burgess R, Iraheta E, Kellerer R, Becherer K, Gessner CR, Li S, Hewitt K, Vuori K, Pasquale EB, et al.: Molecular determinants for interaction of SHEP1 with Cas localize to a highly solvent-protected region in the complex. FEBS Lett 2006, 580(1):175–178. 10.1016/j.febslet.2005.11.070View ArticlePubMedGoogle Scholar
- Black BE, Foltz DR, Chakravarthy S, Luger K, Woods VL, Cleveland DW: Structural determinants for generating centromeric chromatin. Nature 2004, 430(6999):578–582. 10.1038/nature02766View ArticlePubMedGoogle Scholar
- Powell KD, Fitzgerald MC: High-throughput screening assay for the tunable selection of protein ligands. J Combinat Chem 2004, 6(2):262–269. 10.1021/cc034051eView ArticleGoogle Scholar
- Spraggon G, Pantazatos D, Klock HE, Wilson IA, Woods VL, Lesley SA: On the use of DXMS to produce more crystallizable proteins: Structures of the T-maritima proteins TM0160 and TM1171 (vol 13, pg 3187, 2004). Protein Sci 2005, 14(6):1688–1688.Google Scholar
- Wong L, Miyashita O, Woods VL, Onuchic JN, Adams JA, Jennings PA: Protein-protein interactions of Csk probed by H-D exchange and computational analysis. Protein Sci 2004, 13: 201–201.Google Scholar
- Chalmers MJ, Busby SA, Pascal BD, He Y, Hendrickson CL, Marshall AG, Griffin PR: Probing protein ligand interactions by automated hydrogen/deuterium exchange mass spectrometry. Anal Chem 2006, 78(4):1005–1014. 10.1021/ac051294fView ArticlePubMedGoogle Scholar
- Tong Y, Wuebbens MM, Rajagopalan KV, Fitzgerald MC: Thermodynamic analysis of subunit interactions in Escherichia coli molybdopterin synthase. Biochemistry 2005, 44(7):2595–2601. 10.1021/bi047762hView ArticlePubMedGoogle Scholar
- Powell K, Wang M, Silinski P, Ma L, Wales T, Dai S, Warner A, Yang X, Fitzgerald M: The accuracy and precision of a new H/D exchange- and mass spectrometry-based technique for measuring the thermodynamic stability of proteins. Anal Chim Acta 2003, 496(1–2):225–232. 10.1016/S0003-2670(03)01002-XView ArticleGoogle Scholar
- Ma L, Fitzgerald MC: A new H/D exchange- and mass spectrometry-based method for thermodynamic analysis of protein-DNA interactions. Chem Biol 2003, 10(12):1205–1213. 10.1016/j.chembiol.2003.11.017View ArticlePubMedGoogle Scholar
- Powell KD, Fitzgerald MC: Measurements of protein stability by H/D exchange and matrix-assisted laser desorption/ionization mass spectrometry using picomoles of material. Anal Chem 2001, 73(14):3300–3304. 10.1021/ac0100805View ArticlePubMedGoogle Scholar
- Chalmers MJ, Busby SA, Pascal BD, Southern MR, Griffin PR: A two-stage differential hydrogen deuterium exchange method for the rapid characterization of protein/ligand interactions. J Biomol Tech 2007, 18(4):194–204.PubMed CentralPubMedGoogle Scholar
- Weis DD, Engen JR, Kass IJ: Semi-automated data processing of hydrogen exchange mass spectra using HX-Express. J Am Soc Mass Spectrom 2006, 17(12):1700–1703. 10.1016/j.jasms.2006.07.025View ArticlePubMedGoogle Scholar
- Pascal BD, Chalmers MJ, Busby SA, Mader CC, Southern MR, Tsinoremas NF, Griffin PR: The Deuterator: software for the determination of backbone amide deuterium levels from H/D exchange MS data. BMC Bioinformatics 2007, 8: 156. 10.1186/1471-2105-8-156PubMed CentralView ArticlePubMedGoogle Scholar
- Pascal BD, Chalmers MJ, Busby SA, Griffin PR: HD desktop: an integrated platform for the analysis and visualization of H/D exchange data. J Am Soc Mass Spectrom 2009, 20(4):601–610. 10.1016/j.jasms.2008.11.019PubMed CentralView ArticlePubMedGoogle Scholar
- Hotchko M, Anand GS, Komives EA, Ten Eyck LF: Automated extraction of backbone deuteration levels from amide H/2H mass spectrometry experiments. Protein Sci 2006, 15(3):583–601. 10.1110/ps.051774906PubMed CentralView ArticlePubMedGoogle Scholar
- Slysz GW, Baker CA, Bozsa BM, Dang A, Percy AJ, Bennett M, Schriemer DC: Hydra: software for tailored processing of H/D exchange data from MS or tandem MS analyses. BMC Bioinformatics 2009, 10: 162. 10.1186/1471-2105-10-162PubMed CentralView ArticlePubMedGoogle Scholar
- Nikamanon P, Pun E, Chou W, Koter MD, Gershon PD: "TOF2H": a precision toolbox for rapid, high density/high coverage hydrogen-deuterium exchange mass spectrometry via an LC-MALDI approach, covering the data pipeline from spectral acquisition to HDX rate analysis. BMC Bioinformatics 2008, 9: 387. 10.1186/1471-2105-9-387PubMed CentralView ArticlePubMedGoogle Scholar
- Chik JK, Vande Graaf JL, Schriemer DC: Quantitating the statistical distribution of deuterium incorporation to extend the utility of H/D exchange MS data. Anal Chem 2006, 78(1):207–214. 10.1021/ac050988lView ArticlePubMedGoogle Scholar
- Dai SY, Burris TP, Dodge JA, Montrose-Rafizadeh C, Wang Y, Pascal BD, Chalmers MJ, Griffin PR: Unique ligand binding patterns between estrogen receptor alpha and beta revealed by hydrogen-deuterium exchange. Biochemistry 2009, 48(40):9668–9676. 10.1021/bi901149tPubMed CentralView ArticlePubMedGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an open access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.