Search

2D-DIGE is still a very widespread technique in proteomics for the identification of panels of biomarkers, allowing to tackle with some important drawback of classical two-dimensional gel-electrophoresis. However, once 2D-gels are obtained, they must undergo a quite articulated multistep image analysis procedure before the final differential analysis via statistical mono- and multivariate methods. Here, the main steps of image analysis software are described and the most recent procedures reported in the literature are briefly presented.

2-D gel electrophoresis usually provides complex maps characterized by a low reproducibility: this hampers the use of spot volume data for the identification of reliable biomarkers. Under these circumstances, effective and robust methods for the comparison and classification of 2-D maps are fundamental for the identification of an exhaustive panel of candidate biomarkers. Multivariate methods are the most suitable since they take into consideration the relationships between the variables, i.e., effects of synergy and antagonism between the spots. Here the most common multivariate methods used in spot volume datasets analysis are presented. The methods are applied on a sample dataset to prove their effectiveness.

Hierarchical grid transformation is a powerful hierarchical approach to 2-D map warping, able to model both global and local deformations. The algorithm can be stopped when a desired degree of accuracy in the images alignment is obtained. The deformed image is warped and aligned to the target image using a grid where the number of nodes increases in each step of the algorithm. The numerical optimization of the position of the nodes of the grid can be efficiently solved by genetic algorithms, ensuring the achievement of the optimal position of the nodes with a low computational cost with respect to other methods. Here, the optimization of the position of the nodes is carried out by GENOCOP (genetic algorithm for numerical optimization of constrained problems), refined by the following conjugate gradient optimization step. The modeling of the warped space is then achieved by a spline model where some constraints are introduced in the choice of the nodes that are moved. The whole procedure can be intended as an evolutionary method that models the deformation of the gel map at different levels of detail.

The comparison of 2-D maps is not trivial, the main difficulties being the high complexity of the sample and the large experimental variability characterizing 2-D gel electrophoresis. The comparison of maps from control and treated samples is usually performed by specific software, providing the so-called spot volume dataset where each spot of a specific map is matched to its analogous in other maps, and they are described by their optical density, which is supposed to be related to the underlying protein amount. Here, a different approach is presented, based on the direct comparison of 2-D map images: each map is decomposed in terms of moment functions, successively applying the multivariate tools usually adopted in image analysis problems. The moments calculated are then treated with multivariate classification techniques. Here, two types of moment functions are presented (Legendre and Zernike moments), while linear discriminant analysis and partial least squares discriminant analysis are exploited as classification tools to provide the classification of the samples. The procedure is applied to a sample dataset to prove its effectiveness.

Gel electrophoresis is usually applied to identify different protein expression profiles in biological samples (e.g., control vs. pathological, control vs. treated). Information about the effect to be investigated (a pathology, a drug, a ripening effect, etc.) is however generally confounded with experimental variability that is quite large in 2-DE and may arise from small variations in the sample preparation, reagents, sample loading, electrophoretic conditions, staining and image acquisition. Obtaining valid quantitative estimates of protein abundances in each map, before the differential analysis, is therefore fundamental to provide robust candidate biomarkers. Normalization procedures are applied to reduce experimental noise and make the images comparable, improving the accuracy of differential analysis. Certainly, they may deeply influence the final results, and to this respect they have to be applied with care. Here, the most widespread normalization procedures are described both for what regards the applications to 2-DE and 2D Difference Gel-electrophoresis (2-D DIGE) maps.

Two approaches to the analysis of 2-D maps are available: the first one involves a step of spot detection on each gel image; the second one is based instead on the direct differential analysis of 2-D map images, following a pixel-based procedure. Both approaches strongly depend on the proper alignment of the gel images, but the pixel-based approach allows to solve important drawbacks of the spot-volume procedure, i.e., the problem of missing data and of overlapping spots. However, this approach is quite computationally intensive and requires the use of algorithms able to separate the information (i.e., spot-related information) from the background. Here, the most recent pixel-based approaches are described.

Software-based image analysis of 2-D PAGE maps is an important step for the investigation of proteome. Warping algorithms, which are employed to register spots among gels, are able to overcome the difficulties due to the low reproducibility of this analytical technique. Over the years, the research of new matching and warping mathematical methods has allowed the development of several routine applications of easy-to-use software. This chapter describes common and basic spatial transformations used for the alignment of protein spots present in different gel maps; some recently new approaches are also presented.