An improved pixel-based approach for analyzing images in two-dimensional gel electrophoresis

An improved pixel-based approach for analyzing 2-DE images is presented. The key feature of the method is to create a mask based on all gels in the experiment using image morphology, followed by multivariate analysis on the pixel level. The method reduces the impact of noise and background by identifying regions in the image where protein spots are present, but make no assumption on individual spot boundaries for isolated spots. This makes it possible to detect significant changes in complex regions, and visualize these changes over multiple gels in an easy way. False missing values and spot volumes caused by imposing erroneous spot boundaries are thus circumvented. The approach presented gives improved pixel-based information from the gels, and is also an alternative to existing methods for data-reduction, significance testing and visualization of 2-DE data. Results are compared with software using a common spot boundary approach on an experiment consisting of 35 full size gel images. Gel alignment is required before analysis.     

Explorative data analysis of two-dimensional electrophoresis gels

Methods for classification of two-dimensional (2-DE) electrophoresis gels based on multivariate data analysis are demonstrated. Two-dimensional gels of ten wheat varieties are analyzed and it is demonstrated how to classify the wheat varieties in two qualities and a method for initial screening of gels is presented. First, an approach is demonstrated in which no prior knowledge of the separated proteins is used. Alignment of the gels followed by a simple transformation of data makes it possible to analyze the gels in an automated explorative manner by principal component analysis, to determine if the gels should be further analyzed. A more detailed approach is done by analyzing spot volume lists by principal components analysis and partial least square regression. The use of spot volume data offers a mean to investigate the spot pattern and link the classified protein patterns to distinct spots on the gels for further investigation. The explorative approach in analysis of 2-D gels makes it possible, in a fast and convenient way, to screen many gels in order to determine the protein patterns that form clusters and could be selected for further examination.     

Use of principal components analysis for mutation detection with two-dimensional electrophoresis protein separations.

The application of two-dimensional electrophoresis (2-DE) to mutation detection requires the capability to monitor each protein in a 2-DE pattern for significant changes in abundance indicative of a mutation event. Previously, mutation searches were done using a univariate outlier detection method in which each protein spot was considered independently in a classical outlier search. An alternative approach to analysis of 2-DE patterns for quantitative changes is a multivariate procedure which takes advantage of the observation that protein spots in a 2-DE pattern often represent correlated rather than independent measurements. We have compared the efficiency of univariate and multivariate procedures for mutation detection using data from the Argonne National Laboratory 2-DE database of mouse liver proteins. Analyses involving a total of over 1500 gels were performed to compare the performance of a multivariate method based on principal components analysis (PCA) with the univariate method. Up to 279 spots from each pattern were used for PCA. First, a simulation was performed to assess the detection efficiency of PCA for single protein spots decreased in abundance by 50%. Then, the ability to detect actual mutations was tested using eight confirmed mutations. Results show that, compared to a univariate approach to analysis of data from the mouse model system, the multivariate method increases the number of protein spots on each 2-DE pattern that can be monitored for quantitative changes indicative of mutations by compensating for variables that contribute to the background quantitative variability of protein spots.     

The dynamic range of protein expression: a challenge for proteomic research

Proteomic research, for its part, is benefiting enormously from the last decade of genomic research as we now have archived, annotated and audited sequence databases to correlate and query experimental data. While the two-dimensional electrophoresis (2-DE) gels are still a central part of proteomics, we reflect on the possibilities and realities of the current 2-DE technology with regard to displaying and analysing proteomes. Limitations of analysing whole cell/tissue lysates by 2-DE alone are discussed, and we investigate whether extremely narrow p/ranges (1 pH unit/25 cm) provide a solution to display comprehensive protein expression profiles. We are confronted with a challenging task: the dynamic range of protein expression. We believe that most of the existing technology is capable of displaying many more proteins than is currently achievable by integrating existing and new techniques to prefractionate samples prior to 2-DE display or analysis. The availability of a "proteomics toolbox", consisting of defined reagents, methods, and equipment, would assist a comprehensive analysis of defined biological systems.     

Independent component analysis and its applications in signal processing for analytical chemistry

Independent component analysis (ICA) is a statistical method the goal of which is to find a linear representation of non-Gaussian data so that the components are statistically independent, or as independent as possible. In an ICA procedure, the estimated independent components (ICs) are identical to or highly correlated to the spectral profiles of the chemical components in mixtures under certain circumstances, so the latent variables obtained are chemically interpretable and useful for qualitative analysis of mixtures without prior information about the sources or reference materials, and the calculated demixing matrix is useful for simultaneous determination of polycomponents in mixtures. We review commonly used ICA algorithms and recent ICA applications in signal processing for qualitative and quantitative analysis. Furthermore, we also review the preprocessing method for ICA applications and the robustness of different ICA algorithms, and we give the empirical criterion for selection of ICA algorithms in signal processing for analytical chemistry.     

Quantitative and qualitative measure of intralaboratory two-dimensional protein gel reproducibility and the effects of sample preparation, sample load, and image analysis

We investigate one approach to assess the quantitative variability in two-dimensional gel electrophoresis (2-DE) separations based on gel-to-gel variability, sample preparation variability, sample load differences, and the effect of automation on image analysis. We observe that 95% of spots present in three out of four replicate gels exhibit less than a 0.52 coefficient of variation (CV) in fluorescent stain intensity (% volume) for a single sample run on multiple gels. When four parallel sample preparations are performed, this value increases to 0.57. We do not observe any significant change in quantitative value for an increase or decrease in sample load of 30% when using appropriate image analysis variables. Increasing use of automation, while necessary in modern 2-DE experiments, does change the observed level of quantitative and qualitative variability among replicate gels. The number of spots that change qualitatively for a single sample run in parallel varies from a CV = 0.03 for fully manual analysis to CV = 0.20 for a fully automated analysis. We present a systematic method by which a single laboratory can measure gel-to-gel variability using only three gel runs.     

Analysis of high molecular mass proteins larger than 150 kDa using cyanogen bromide cleavage and conventional 2-DE

Proteomic approaches including high-resolution 2-DE are providing the tools needed to discover disease-associated biomarkers in complex biological samples. Although 2-DE is an extremely powerful approach to analyze the proteome, the separation of proteins with extreme molecular masses still remains an issue requiring improvement. Because high molecular mass (HMM) proteins larger than 150 kDa have already been observed to be differentially expressed in several pathologies such as cancer, we developed an original strategy to analyze this part of the proteome that is not easily separated by 2-DE in polyacrylamide gels. This strategy is based on the 2-DE separation of cyanogen bromide (CNBr) fragments of purified HMM protein fractions, and combines techniques including SEC fractionation, TCA precipitation, CNBr cleavage, 2-DE and MS analysis. The method was first tested on a model protein, the BSA. Preliminary results obtained using colonic tissues led to the identification of six HMM proteins with M(r) comprised between 163 and 533 kDa in their reduced state. These results demonstrated that our CNBr/2-DE approach should provide a powerful tool for identification of new biomarkers larger than 150 kDa.     

Establishment of “one-piece” large-gel 2-DE for high-resolution analysis of small amounts of sample using difference gel electrophoresis saturation labelling

Large-gel two-dimensional gel electrophoresis (2-DE) is the method of choice for high-resolution proteome analysis of complex protein mixtures. Until now, however, the advantages of large 2-DE in combination with multiplexed fluorescence dye protein labelling has been complicated by the separate handling and analysis of the second-dimension gels. Therefore, we adapted the large 2-DE procedure allowing us to run “one-piece” large 2-DE gels (40 cm × 30 cm) in the second dimension for high resolution proteome analysis. Here, we show that in combination with fluorescence dye protein saturation labelling “one-piece” large 2-DE enables analysis of small amounts of sample (3 μg protein) for high-resolution proteome analysis.     

A protein extraction method compatible with proteomic analysis for the euhalophyte Salicornia europaea

Protein extraction from plants like the halophyte Salicornia europaea has been problematic using standard protocols due to high concentrations of salt ions in their cells. We have developed an improved method for protein extraction from S. europaea, which allowed us to remove interfering compounds and salt ions by including the chemicals borax, polyvinylpolypyrrolidone, and phenol. The comparative study of this method with several other protocols using NaCl-treated S. europaea shoots demonstrated that this method gave the best distinction of proteins on 2-DE gels. This protocol had a wide range of applications as high yields and good distinction of 1-DE gels for proteins isolated from twelve other plants were rendered. In addition, we reported results of 2-DE using the recalcitrant tissue of the S. europaea roots. We also demonstrated that this protocol is compatible with proteomic analysis as eight specific proteins generated by this method have been identified by MS. In conclusion, our newly developed protein extraction protocol is expected to have excellent applications in proteomic studies of halophytes.     

Sample complexity reduction for two-dimensional electrophoresis using solution isoelectric focusing prefractionation

Despite its excellent resolving power, 2-DE is of limited use when analyzing cellular proteomes, especially in differential expression studies. Frequently, fewer than 2000 protein spots are detected on a single 2-D gel (a fraction of the total proteome) regardless of the gel platform, sample, or detection method used. This is due to the vast number of proteins expressed and their equally vast dynamic range. To exploit 2-DE unique ability as both an analytical and a preparative tool, the significant sample prefractionation is necessary. We have used solution isoelectric focusing (sIEF) via the ZOOM IEF Fractionator (Invitrogen) to generate sample fractions from complex bacterial lysates, followed by parallel 2-DE, using narrow-range IPG strips that bracket the sIEF fractions. The net result of this process is a significant enrichment of the bacterial proteome resolved on multiple 2-D gels. After prefractionation, we detected 5525 spots, an approximate 3.5-fold increase over the 1577 spots detected in an unfractionated gel. We concluded that sIEF is an effective means of prefractionation to increase depth of field and improve the analysis of low-abundance proteins.     

Comments on selected fundamental aspects of microarray analysis

Microarrays are becoming a ubiquitous tool of research in life sciences. However, the working principles of microarray-based methodologies are often misunderstood or apparently ignored by the researchers who actually perform and interpret experiments. This in turn seems to lead to a common over-expectation regarding the explanatory and/or knowledge-generating power of microarray analyses. In this note we intend to explain basic principles of five (5) major groups of analytical techniques used in studies of microarray data and their interpretation: the principal component analysis (PCA), the independent component analysis (ICA), the t-test, the analysis of variance (ANOVA), and self organizing maps (SOM). We discuss answers to selected practical questions related to the analysis of microarray data. We also take a closer look at the experimental setup and the rules, which have to be observed in order to exploit microarrays efficiently. Finally, we discuss in detail the scope and limitations of microarray-based methods. We emphasize the fact that no amount of statistical analysis can compensate for (or replace) a well thought through experimental setup. We conclude that microarrays are indeed useful tools in life sciences but by no means should they be expected to generate complete answers to complex biological questions. We argue that even well posed questions, formulated within a microarray-specific terminology, cannot be completely answered with the use of microarray analyses alone.     

Independent component analysis yields chemically interpretable latent variables in multivariate regression

This work shows that independent component analysis (ICA) can be used to obtain statistically independent and, therefore, chemically interpretable latent variables (LVs) in multivariate regression. Two novel algorithms based on ICA are introduced and compared with two classical methods on simulated data: principal component regression and partial least-squares regression. All methods compared yield accurate predictions, but only those based on ICA yield LVs that are chemically interpretable. Practical limitations of ICA-based regression with respect to the underlying assumptions, sample size, and measurement noise are discussed and illustrated by means of simulations.     

Classification for high-throughput data with an optimal subset of principal components

High-throughput data have been widely used in biological and medical studies to discover gene and protein functions. Due to the high dimensionality, principal component analysis (PCA) is often involved for data dimension reduction. However, when a few principal components (PCs) are selected for dimension reduction or considered for dimension determination, they are typically ranked by their variances, eigenvalues. However, this approach is not always effective in subsequent multivariate analysis, particularly classification. To maximize information from data with a subset of the components, we apply a different ranking criterion, canonical variate criterion, which considers within- and between-group variance rather than total variance in the classical criterion. Four prevalent classification methods are considered and compared using leave-one-out cross-validation. These methods are illustrated with three real high-throughput data sets, two microarray data sets and a nuclear magnetic resonance spectra data set.     

Multivariate analysis of 2-DE protein patterns--practical approaches

Practical approaches to the use of multivariate data analysis of 2-DE protein patterns are demonstrated by three independent strategies for the image analysis and the multivariate analysis on the same set of 2-DE data. Four wheat varieties were selected on the basis of their baking quality. Two of the varieties were of strong baking quality and hard wheat kernel and two were of weak baking quality and soft kernel. Gliadins at different stages of grain development were analyzed by the application of multivariate data analysis on images of 2-DEs. Patterns related to the wheat varieties, harvest times and quality were detected on images of 2-DE protein patterns for all the three strategies. The use of the multivariate methods was evaluated in the alignment and matching procedures of 2-DE gels. All the three strategies were able to discriminate the samples according to quality, harvest time and variety, although different subsets of protein spots were selected. The explorative approach of using multivariate data analysis and variable selection in the analyses of 2-DEs seems to be promising as a fast, reliable and convenient way of screening and transforming many gel images into spot quantities.     

Cross validation and uncertainty estimates in independent component analysis

A data analysis tool, known as independent component analysis (ICA), is the main focus of this paper. The theory of ICA is briefly reviewed, and the underlying statistical assumptions and a practical algorithm are described. This paper introduces cross validation/jack-knifing and significance tests to ICA. Jack-knifing is applied to estimate uncertainties for the ICA loadings, which also serve as a basis for significance tests. These tests are shown to improve ICA performance, indicating how many components are mixed in the observed data, and also which parts of the extracted sources that contain significant information. We address the issue of stability for the ICA model through uncertainty plots. The ICA performance is compared to principal component analysis (PCA) for two selected applications, a simulated experiment and a real world application.     

A multivariate spot filtering model for two-dimensional gel electrophoresis

Image segmentation plays an important role in the automatic analysis of protein spots in 2-DE. Using image segments representing protein spots, the amount of protein in each segment can be quantified, and corresponding segments can be matched and compared for multiple gels. However, the common presence of image segments caused by noise and unwanted artefacts highly disturb the analysis and comparison of the gels. Common sources of such artefacts are cracks in the gel surface, fingerprints, dust and other pollutions. It would be advantageous to remove these unwanted artefacts during or after the segmentation procedure. To achieve this task a multivariate spot filtering model is developed using image segments from a gel segmentation. Parameters in the model are based on texture, shape and intensity measurements of the image segments. The model successfully managed to separate segments caused by noise, artefacts and cracks from image segments representing true protein spots. The classification method used is discriminant partial least squares regression.     

High reproducibility of large-gel two-dimensional electrophoresis

Two-dimensional gel electrophoresis (2-DE) facilitates the separation of thousands of proteins from highly complex protein mixtures and has become a central method in proteomics in recent years. In the present study, we examined the technical variability of large 2-DE gels with respect to sample preparation, electrophoresis procedure, data acquisition, and biological variation by analyzing a disease (Huntington's disease) and control state with a commercially available software package, PROTEOMWEAVER trade mark. Scatter plots and correlation coefficients were obtained to quantify both technical and biological variation. Even 2-DE gels run separately in both dimensions yielded correlation coefficients around 0.88 and deviations from the mean close to 20% for low-intensity spots. This indicates a high technical reproducibility of the 2-DE procedure developed in our laboratory. Variability within a biological condition was low and comparable to technical variation (at least 0.87). Two-dimensional (2-D) gels obtained from samples of different biological conditions (health vs. disease) achieved a variability similar to intracondition and technical variability. These findings highlight the importance of multiple gel and spot-by-spot comparisons to identify biological significant changes. Minor errors introduced by technical and biological variation allow a comparison of all gels within a study which facilitates the tackling of complex biological problems.     

Measurement of specific radioactivity in proteins separated by two-dimensional gel electrophoresis

We report a method to quantify the specific radioactivity of proteins that have been separated by 2-DE. Gels are stained with SyproRuby, and protein spots are excised. The SyproRuby dye is extracted from each spot using DMSO, and the fluorescence is quantified automatically using a plate reader. The extracted gel piece is then dissolved in hydrogen peroxide and radioactivity is quantified by liquid scintillation counting. Gentle agitation with DMSO for 24 h was found to extract all the SyproRuby dye from gel fragments. The fluorescence of the extract was linearly related to the amount of BSA loaded onto a series of 1-D gels. When rat muscle samples were run on 2-DE gels, the fluorescence extracted from 54 protein spots showed a good correlation (r = 0.79, p < 0.001) with the corresponding spot intensity measured by conventional scanning and image analysis. DMSO extraction was found not to affect the amount of radioactive protein left in the gel. When a series of BSA solutions of known specific radioactivity were run on 2-DE gels, the specific radioactivity measured by the new method showed a good correlation (r = 0.98, p < 0.01, n = 5) with the specific radioactivity measured directly before loading. Reproducibility of the method was measured in a series of 2-DE gels containing proteins from the livers of rats and mice that had been injected with [35S]methionine. Variability tended to increase when the amount of radioactivity in the protein spot was low, but for samples containing at least 10 dpm above background the CV was around 30%, which is comparable to that obtained when measuring protein expression by conventional image analysis of SyproRuby-stained 2-DE gels. Similar results were obtained whether spots were excised manually or using a spot excision robot. This method offers a high-throughput, cost-effective and reliable method of quantifying the specific radioactivity of proteins from metabolic labelling experiments carried out in vivo, so long as sufficient quantities of radioactive tracer are used.     

Software-induced variance in two-dimensional gel electrophoresis image analysis

Experimental variability in 2-DE is well documented, but little attention has been paid to variability arising from postexperimental quantitative analyses using various 2-DE software packages. The performance of two 2-DE analysis software programs, Phoretix 2D Expression v2004 (Expression) and PDQuest 7.2 (PDQuest), was evaluated in this study. All available background subtraction and smoothing algorithms were tested using both data generated from one single 2-DE gel image, thus excluding experimental variance, and with authentic sets of replicate gels (n = 5). A slight shift of the image boundaries (the "cropping area") caused both programs to induce variance in protein spot quantification of otherwise identical gel images. The resulting variance for PDQuest (CV(mean) = 8%) was approximately twice that for Expression (CV(mean) = 4%). In authentic sets of replicate 2-DE gels (n = 5), the experimental variance confounded the software-induced variance to some extent. However, Expression still outperformed PDQuest, which exhibited software-induced variance as high as 25% of the total observed variance. Surprisingly, the complete omission of background subtraction algorithms resulted in the least amount of software-based variance. These data indicate that 2-DE gel analysis software constitutes a significant source of the variance observed in quantitative proteomics, and that the use of background subtraction algorithms can further increase the variance.