Investigation of Segmentation Methods for Proteomics

Protein spot identification in two-dimensional electrophoresis gels can be supported by the comparison of gel images accessible in different World Wide Web two-dimensional electrophoresis (2-DE) gel protein databases. The comparison may be performed either by visual cross-matching between gel images or by automatic recognition of similar protein spot patterns. A prerequisite for the automatic point pattern matching approach is the detection of protein spots yielding the x(s),y(s) coordinates and integrated spot intensities i(s). For this purpose an algorithm is developed based on a combination of hierarchical watershed transformation and feature extraction methods. This approach reduces the strong over-segmentation of spot regions normally produced by watershed transformation. Measures for the ellipticity and curvature are determined as features of spot regions. The resulting spot lists containing x(s),y(s),i(s)-triplets are calculated for a source as well as for a target gel image accessible in 2-DE gel protein databases. After spot detection a matching procedure is applied. Both the matching of a local pattern vs. a full 2-DE gel image and the global matching between full images are discussed. Preset slope and length tolerances of pattern edges serve as matching criteria. The local matching algorithm relies on a data structure derived from the incremental Delaunay triangulation of a point set and a two-step hashing technique. For the incremental construction of triangles the spot intensities are considered in decreasing order. The algorithm needs neither landmarks nor an a priori image alignment. A graphical user interface for spot detection and gel matching is written in the Java programming language for the Internet. The software package called CAROL (http://gelmatching.inf.fu-berlin.de) is realized in a client-server architecture.

In this paper, a novel method to automatically detect protein spots on a two-dimensional (2-D) electrophoresis gel image is proposed to implement proteomics analysis of complex analyte. On the basis of the identifying spots results based on color variation and spot size features, morphological feature is introduced as a new criterion to distinguish protein spots from non-protein spots. Image-sharpening, edge-detecting and morphological feature extraction methods were consequently combined to detect protein spots on a 2-D electrophoresis gel image subject to strong disturbance. The proposed method was applied to detect the protein spots of proteomic gel images from E. coli cell, human kidney tissue and human serum. The results demonstrated that this method is more accurate and reliable than previous methods such as PDQuest 7.2 and ImageMaster 5.0 software for detecting protein spots on gel images with strong interferences.

The aim of the investigation presented in this paper was to develop a software-based assistant for the protein analysis workflow. The prior characterization of the unknown protein in two-dimensional electrophoresis gel images is performed according to the molecular weight and isoelectric point of each protein spot estimated from the gel image before further sequence analysis by mass spectrometry. The paper presents a method for automatic and robust identification of the protein standard band in a two-dimensional gel image. In addition, the method introduces the identification of the positions of the markers, prepared by using pre-selected proteins with known molecular mass. The robustness of the method was achieved by using special validation rules in the proposed original algorithms. In addition, a self-organizing map-based decision support algorithm is proposed, which takes Gabor coefficients as image features and searches for the differences in preselected vertical image bars. The experimental investigation proved the good performance of the new algorithms included into the proposed method. The detection of the protein standard markers works without modification of algorithm parameters on two-dimensional gel images obtained by using different staining and destaining procedures, which results in different average levels of intensity in the images.

The cell wall proteins of Candida albicans play a key role in morphogenesis and pathogenesis and might be potential target sites for new specific antifungal drugs. However, these proteins are difficult to analyze because of their high heterogeneity, interconnections with wall polysaccharides (mannan, glucan, and chitin), low abundance, low solubility, and hydrophobic nature. Here we report a subproteomic approach for the study of the cell wall proteins (CWPs) from C. albicans yeast and hyphal forms. Most of the mannoproteins present in this compartment were extracted by cell wall fractionation according to the type of interactions that they establish with other structural components. CWPs were solubilized from isolated cell walls by hot SDS and dithiothreitol treatment followed by extraction either by mild alkali conditions or by enzymatic treatment with glucanases and chitinases. These highly enriched cell wall fractions were analyzed by two-dimensional PAGE, showing that a large number of proteins are involved in cell wall construction and that the wall remodeling that occurs during germ tube formation is related to changes in the composition of CWPs. We suggest that the CWP-chitin linkage is an important retention mechanism of CWPs in C. albicans mycelial forms. This article also highlights the usefulness of the combination of sequential fractionation and two-dimensional PAGE followed by Western blotting using specific antibodies against known CWPs in the characterization of incorporation mechanisms of such CWPs into the cell wall and of their interactions with other wall components. Mass spectrometry analyses have allowed the identification of several cell surface proteins classically associated with both the cell wall and other compartments. The physiological significance of the dual location of these moonlighting proteins is also discussed. This approach is therefore a powerful tool for obtaining a comprehensive and integrated view of the cell wall proteome.

BackgroundHistopathology image analysis is a gold standard for cancer recognition and diagnosis. Automatic analysis of histopathology images can help pathologists diagnose tumor and cancer subtypes, alleviating the workload of pathologists. There are two basic types of tasks in digital histopathology image analysis: image classification and image segmentation. Typical problems with histopathology images that hamper automatic analysis include complex clinical representations, limited quantities of training images in a dataset, and the extremely large size of singular images (usually up to gigapixels). The property of extremely large size for a single image also makes a histopathology image dataset be considered large-scale, even if the number of images in the dataset is limited.ResultsIn this paper, we propose leveraging deep convolutional neural network (CNN) activation features to perform classification, segmentation and visualization in large-scale tissue histopathology images. Our framework transfers features extracted from CNNs trained by a large natural image database, ImageNet, to histopathology images. We also explore the characteristics of CNN features by visualizing the response of individual neuron components in the last hidden layer. Some of these characteristics reveal biological insights that have been verified by pathologists. According to our experiments, the framework proposed has shown state-of-the-art performance on a brain tumor dataset from the MICCAI 2014 Brain Tumor Digital Pathology Challenge and a colon cancer histopathology image dataset.ConclusionsThe framework proposed is a simple, efficient and effective system for histopathology image automatic analysis. We successfully transfer ImageNet knowledge as deep convolutional activation features to the classification and segmentation of histopathology images with little training data. CNN features are significantly more powerful than expert-designed features.

Alignment of two-dimensional electrophoresis gels

Two-dimensional gel electrophoresis (2DE) is an indispensable tool in proteomics for the analysis of protein expression in complex biological systems such as cells and tissues. However, the automatic extraction of information from gel images is still a challenging task. In this paper we propose a strategy that represents a computational procedure of support to the discrimination of different clinical conditions associated with the samples. The analyzed gel images were acquired within the framework of a study of peripheral neuropathies: twenty-four 2DE maps generated from cerebrospinal fluid (16 pathologic and 8 control subjects) were processed. Quantitative features were defined to describe each image and treated with a method of dimensionality reduction. The informativeness of the descriptors allowed us to see the gel of the data set as items in a three-dimensional space, segregating according to the clinical conditions. Moreover, information with prognostic value was obtained for a single outsider gel of a patient who was included in a clinical subgroup at the first diagnosis but whose disease progressed with clinical features belonging to a different clinical subgroup. The method developed may represent an effective tool of classification that can be used repeatedly to capture the essential impression from separation images.

In previous studies, two-dimensional polyacrylamide gel electrophoresis allowed the detection of 11 polypeptide markers that distinguished between subtypes of acute lymphoblastic leukemia and between acute lymphoblastic and acute myelocytic leukemia. This report describes the occurrence of polypeptide markers in two-dimensional gels that indicate a myeloid origin of blasts obtained from two children who presented with acute leukemia, the cells of origin of which could not be determined, at the time of diagnosis, by morphologic, cytochemical, or immune marker analysis. The authors conclude that two-dimensional electrophoresis provides a new tool for the delineation of the cell of origin in acute leukemia.

Two-dimensional gel electrophoresis (2-DE) is the most established protein separation method used in expression proteomics. Despite the existence of sophisticated software tools, 2-DE gel image analysis still remains a serious bottleneck. The low accuracies of commercial software packages and the extensive manual calibration that they often require for acceptable results show that we are far from achieving the goal of a fully automated and reliable, high-throughput gel processing system. We present a novel spot detection and quantification methodology which draws heavily from unsupervised machine-learning methods. Using the proposed hierarchical machine learning-based segmentation methodology reduces both the number of faint spots missed (improves sensitivity) and the number of extraneous spots introduced (improves precision). The detection and quantification performance has been thoroughly evaluated and is shown to compare favorably (higher F-measure) to a commercially available software package (PDQuest). The whole image analysis pipeline that we have developed is fully automated and can be used for high-throughput proteomics analysis since it does not require any manual intervention for recalibration every time a new 2-DE gel image is to be analyzed. Furthermore, it can be easily parallelized for high performance and also applied without any modification to prealigned group average gels.

Two-dimensional gel electrophoresis (2-DE) remains the method of choice for the Separation of protein mixtures whilst mass spectrometry (MS) is rapidly becoming the premier tool for protein identification. When combined, 2-DE and MS form the current operating paradigm for classical proteomics. One of the key challenges of proteome research is that of detecting and identifying all of the elements (proteins) of a proteome. Silver staining and radiolabelling, e.g. with 35S-methionine ([35S]-met), represent two sensitive methods used to visualise many of the constitutive and synthesised elements of a proteome, respectively. The latter method allows a very low total protein loading on a two-dimensional (2-D) gel and challenges protein identification using current MS-based technology. Therefore, it is necessary to refer to and locate a radiolabelled spot's cognate on a preparatively loaded stained gel, or Western blot, and use that protein spot for identification. Unfortunately, the images of autoradiographs and preparative gels or blots, even of the same sample, often do not correspond making it difficult to accurately locate and select spots of interest by visual comparison. We have established a technique that permits the unambiguous localisation of radiolabelled proteins on the same silver stained 2-D gel. Protein identification of superimposed spots is described by peptide mass fingerprinting and database searching using matrix-assisted laser desorption/ionization-time of flight mass spectrometry and by peptide sequencing using tandem MS by hybrid quadrupole/orthogonal acceleration time of flight MS (Q-TOF).

The two-dimensional electrophoresis (2-DE) method is widely used in proteomics to separate thousands of proteins in a sample. Given a set of gel images of the same kind, a spot-matching operation identifies the spot of the same protein in each 2-DE gel image. However, since 2-DE gel images are very sensitive to various experimental conditions, spots observed in a 2-DE gel image involve inherent errors such as noises and irregular geometric distortions. Consequently, the accuracy of spot-matching is intensively influenced by these errors. This paper proposes a spot-matching framework to improve the accuracy of 2-DE gel image analysis. The framework includes an automatic landmark extraction module that uses a pattern model of positional and intensive distortions identified in the learning set of gel images. Subsequently, a probabilistic matching adjustment module is employed to produce more reliable spot-matching results based on the selected landmark spots in the first module. In order to confirm the synergy effects of integrating these two modules, the improved accuracy of the proposed framework is analyzed through various experiments on real 2-DE gel images of human liver tissue.

DNA electrophoresis gel is an important biologically experimental technique and DNA sequencing can be defined by it. Traditionally, it is time consuming for biologists to exam the gel images by their eyes and often has human errors during the process. Therefore, automatic analysis of the gel image could provide more information that is usually ignored by human expert. However, basic tasks such as the identification of lanes in a gel image, easily done by human experts, emerge as problems that may be difficult to be executed automatically. In this paper, we design an automatic procedure to analyze DNA gel images using various image processing algorithms. Firstly, we employ an enhanced fuzzy c-means algorithm to extract the useful information from DNA gel images and exclude the undesired background. Then, Gaussian function is utilized to estimate the location of each lane of A, T, C, and G on the gels images automatically. Finally, the location of each band on the gel image can be detected accurately by tracing lanes, renewing lost bands, and eliminating repetitive bands.

In this paper, we focus on the quantitative comparison of two-dimensional gel images in order to indicate the possible changes of the proteome in separate samples. The comparison of two or more separate gel images cannot be straightforward because it is hard to ensure the same experimental conditions for each experiment. The aim of our investigation was to increase the precision of quantitative validation of changes in proteome of analyzed biological tissue. Our work was divided into two parts: analysis of uncertainties during quantification of gel images prepared with different staining intensities, images with saturated protein spots and to propose an image processing technique to minimize these uncertainties. We proposed a SOM based clustering of watershed transformation based image segmentation result and separate comparison of protein spot intensities in each group. The experimental investigation showed, that the effective SOM size is 3×3 and background subtraction based on pixel intensities on image segment contour give less uncertainty comparing to alternative techniques.

Back ground: Adipose tissue plays a crucial role in regulation of lipid and carbohydrate metabolism and many of these actions are regulated by insulin. Objective: The present study aims to detect potential novel protein targets for insulin sensitization in human adipose tissue by using an unbiased proteomic approaches. Method: Ten moderately obese, but otherwise healthy, subjects were treated with rosiglitazone 4 mg BD for 14 days. Protein profiles were obtained by two-dimensional gel (2-DG) electrophoresis and protein spots were identified using tandem mass spectrometry. Data were analyzed statistically by using student's paired t test. Results: Proteomic analysis revealed distinct up or down regulation (≥ 2-fold) in 122 protein spots on the 2-D gel images between day 0 and day 14 adipose tissue samples. Treatment with rosiglitazone increased the expression of extracellular matrix proteins including collagens, galectin-1, nidogen-1 and laminin. Analysis revealed differential expression of several other proteins involved in redox regulation (ferritin light chain, heat-shock protein β-5, heatshock protein β-1), endoplasmic reticulum (ER) stress (endoplasmin, 78 kDa glucose-regulated protein precursor), cytoskeletal reorganization (lamin A/C, vimentin, tropomyosin-3, tropomyosin-4, gelsolin, actin tubulin β-1 chain), signaling (calmodulin), and others (glyceraldehyde-3-phosphate dehydrogenase, α enolase). Conclusion: Such orchestrated changes in expression of multiple proteins provide insights into the mechanism undarlying the increased efficiency in adipocyte remodeling and differentiation and hence improvement of insulin sensitivity in response to rosiglitazone treatment. Key words: Rosiglitazone, adipose tissue, proteomics, collagen, adipogenesis, insulin, electrophoresis, spectrometry. DOI: 10.3329/jbsp.v4i2.4170 J Bangladesh Soc Physiol. 2009 Dec;4(2): 34-43

(1) When rat liver 40 S ribosomal proteins in 6 M urea were were mixed with poly(U) at an appropriate ratio, a precipitate was formed which was also insoluble in the sample solution for two-dimensional acrylamide gel electrophoresis. Analyses by two-dimensional acrylamide gel electrophoresis showed that S7 and S10 proteins (according to our numbering system) had disappeared selectively from the fraction soluble in 6 M urea. These two proteins were present in the fraction insoluble in 6 M urea, and became soluble in the sample solution after treating it with RNase. The results suggest that S7 and S10 proteins have strong affinities for poly(U). When rat liver 40 S subunits were incubated with poly(U), similar results were obtained. (2) After incubation of 40 S subunits with [3H]poly(U) and then with unlabeled poly(U), UV irradiation cross-linked poly(U) to the protein moiety of the 40 S subunit. When the protein fraction insoluble in the sample solution for two-dimensional electrophoresis was prepared from 40 S subunits cross-linked to poly(U) and then subjected to two-dimensional acrylamide gel electrophoresis after RNase treatment, S7 and S10 proteins were detected on the gel. In addition to the S7 protein spot, a triangular area spreading from the spot to the origin contained radioactivity. The results suggest that poly(U) is cross-linked to S7 protein and oligo(U) fragments bound to S7 protein affect its electrophoretic mobility. (3) Ribosomal proteins were prepared from 40 S subunits cross-linked to carrier-free [3H]poly(U) and analyzed by three-dimensional acrylamide gel electrophoresis (Terao, K. & Ogata, K. (1975) Biochim. Biophys. Acta 402, 214--229) after RNase treatment. It was found that S7, S6, and S15 proteins are cross-linked to poly(U). From the results of the present and preceding experiments it is concluded that S7 is the poly(U)-binding protein. The possibility that other proteins in 40 S ribosomal subunits interact with poly(U) is discussed.