Enhanced analytical power of SDS-PAGE using machine learning algorithms

Two-dimensional gel electrophoresis (2D-GE) image analysis based on CellProfiler: Pseudomonas aeruginosa AG1 as model

Two-dimensional gel electrophoresis (2D-GE) is an indispensable technique for the study of proteomes of biological systems, providing an assessment of changes in protein abundance under various experimental conditions. However, due to the complexity of 2D-GE gels, there is no systematic, automatic, and reproducible protocol for image analysis and specific implementations are required for each context. In addition, practically all available solutions are commercial, which implies high cost and little flexibility to modulate the parameters of the algorithms. Using the bacterial strain, Pseudomonas aeruginosaAG1 as a model, we obtained images from 2D-GE of periplasmic protein profiles when the strain was exposed to multiple conditions, including antibiotics. Then, we proceeded to implement and evaluate an image analysis protocol with open-source software, CellProfiler. First, a preprocessing step included a bUnwarpJ-Image pipeline for aligning 2D-GE images. Then, using CellProfiler, we standardized two pipelines for spots identification. Total spots recognition was achieved using segmentation by intensity, whose performance was evaluated when compared with a reference protocol. In a second pipeline with the same program, differential identification of spots was addressed when comparing pairs of protein profiles. Due to the characteristics of the programs used, our workflow can automatically analyze a large number of images and it is parallelizable, which is an advantage with respect to other implementations. Finally, we compared six experimental conditions of bacterial strain in the presence or absence of antibiotics, determining protein profiles relationships by applying clustering algorithms PCA (Principal Components Analysis) and HC (Hierarchical Clustering).

Rapid Prediction of Bacterial Heterotrophic Fluxomics Using Machine Learning and Constraint Programming

13C metabolic flux analysis (13C-MFA) has been widely used to measure in vivo enzyme reaction rates (i.e., metabolic flux) in microorganisms. Mining the relationship between environmental and genetic factors and metabolic fluxes hidden in existing fluxomic data will lead to predictive models that can significantly accelerate flux quantification. In this paper, we present a web-based platform MFlux (http://mflux.org) that predicts the bacterial central metabolism via machine learning, leveraging data from approximately 100 13C-MFA papers on heterotrophic bacterial metabolisms. Three machine learning methods, namely Support Vector Machine (SVM), k-Nearest Neighbors (k-NN), and Decision Tree, were employed to study the sophisticated relationship between influential factors and metabolic fluxes. We performed a grid search of the best parameter set for each algorithm and verified their performance through 10-fold cross validations. SVM yields the highest accuracy among all three algorithms. Further, we employed quadratic programming to adjust flux profiles to satisfy stoichiometric constraints. Multiple case studies have shown that MFlux can reasonably predict fluxomes as a function of bacterial species, substrate types, growth rate, oxygen conditions, and cultivation methods. Due to the interest of studying model organism under particular carbon sources, bias of fluxome in the dataset may limit the applicability of machine learning models. This problem can be resolved after more papers on 13C-MFA are published for non-model species.

(Post-)genomics approaches in fungal research

To date, hundreds of fungal genomes have been sequenced and many more are in progress. This wealth of genomic information has provided new directions to study fungal biodiversity. However, to further dissect and understand the complicated biological mechanisms involved in fungal life styles, functional studies beyond genomes are required. Thanks to the developments of current -omics techniques, it is possible to produce large amounts of fungal functional data in a high-throughput fashion (e.g. transcriptome, proteome, etc.). The increasing ease of creating -omics data has also created a major challenge for downstream data handling and analysis. Numerous databases, tools and software have been created to meet this challenge. Facing such a richness of techniques and information, hereby we provide a brief roadmap on current wet-lab and bioinformatics approaches to study functional genomics in fungi.

Proteotyping of Holm oak (Quercus ilex subsp. ballota) provenances through proteomic analysis of acorn flour

Proteomics has become a powerful tool to characterize biodiversity and natural variability in plant species, as well as to catalogue and establish phylogenetic relationships and distances among populations, provenances or ecotypes. In this chapter, we describe the standard proteomics workflow that we currently use in cataloguing Holm oak (Quercus ilex subsp. ballota [Desf.] Samp.) populations. Proteins are extracted from acorn flour or pollen by TCA/acetone or TCA/acetone-phenol methods, resolved by one- or two-dimensional gel electrophoresis, and gel images are captured and analyzed by appropriate software and statistical packages. Quantitative or qualitative variable bands or spots are subjected to MS analysis in order to identify them and correlate differences in the protein profile with the phenotypes or environmental conditions.

Making a protein extract from plant pathogenic fungi for gel- and LC-based proteomics

Proteomic technologies have become a successful tool to provide relevant information on fungal biology. In the case of plant pathogenic fungi, this approach would allow a deeper knowledge of the interaction and the biological cycle of the pathogen, as well as the identification of pathogenicity and virulence factors. These two elements open up new possibilities for crop disease diagnosis and environment-friendly crop protection. Phytopathogenic fungi, due to its particular cellular characteristics, can be considered as a recalcitrant biological material, which makes it difficult to obtain quality protein samples for proteomic analysis. This chapter focuses on protein extraction for gel- and LC-based proteomics with specific protocols of our current research with Botrytis cinerea.

White matter differences between healthy young ApoE4 carriers and non-carriers identified with tractography and support vector machines

The apolipoprotein E4 (ApoE4) is an established risk factor for Alzheimer's disease (AD). Previous work has shown that this allele is associated with functional (fMRI) changes as well structural grey matter (GM) changes in healthy young, middle-aged and older subjects. Here, we assess the diffusion characteristics and the white matter (WM) tracts of healthy young (20-38 years) ApoE4 carriers and non-carriers. No significant differences in diffusion indices were found between young carriers (ApoE4+) and non-carriers (ApoE4-). There were also no significant differences between the groups in terms of normalised GM or WM volume. A feature selection algorithm (ReliefF) was used to select the most salient voxels from the diffusion data for subsequent classification with support vector machines (SVMs). SVMs were capable of classifying ApoE4 carrier and non-carrier groups with an extremely high level of accuracy. The top 500 voxels selected by ReliefF were then used as seeds for tractography which identified a WM network that included regions of the parietal lobe, the cingulum bundle and the dorsolateral frontal lobe. There was a non-significant decrease in volume of this WM network in the ApoE4 carrier group. Our results indicate that there are subtle WM differences between healthy young ApoE4 carriers and non-carriers and that the WM network identified may be particularly vulnerable to further degeneration in ApoE4 carriers as they enter middle and old age.

Integrative systems biology: an attempt to describe a simple weed

Genome-scale studies hold great promise for revealing novel plant biology. Because of the complexity of these techniques, numerous considerations need to be made before embarking on a study. Here we focus on the Arabidopsis model system because of the wealth of available genome-scale data. Many approaches are available that provide genome-scale information regarding the state of a given organism (e.g. genomics, epigenomics, transcriptomics, proteomics, metabolomics interactomics, ionomics, phenomics, etc.). Integration of all of these types of data will be necessary for a comprehensive description of Arabidopsis. In this review we propose that 'triangulation' among transcriptomics, proteomics and metabolomics is a meaningful approach for beginning this integrative analysis and uncovering a systems level perspective of Arabidopsis biology.

Isoelectric point optimization using peptide descriptors and support vector machines

IPG (Immobilized pH Gradient) based separations are frequently used as the first step in shotgun proteomics methods; it yields an increase in both the dynamic range and resolution of peptide separation prior to the LC-MS analysis. Experimental isoelectric point (pI) values can improve peptide identifications in conjunction with MS/MS information. Thus, accurate estimation of the pI value based on the amino acid sequence becomes critical to perform these kinds of experiments. Nowadays, pI is commonly predicted using the charge-state model [1], and/or the cofactor algorithm [2]. However, none of these methods is capable of calculating the pI value for basic peptides accurately. In this manuscript, we present an new approach that can significant improve the pI estimation, by using Support Vector Machines (SVM) [3], an experimental amino acid descriptor taken from the AAIndex database [4] and the isoelectric point predicted by the charge-state model. Our results have shown a strong correlation (R(2)=0.98) between the predicted and observed values, with a standard deviation of 0.32 pH units across the complete pH range.

Multivariate evaluation of 1-dimensional sarcoplasmic protein profile patterns of turkey breast muscle during early post-hatch development

Proteins are the main participants in metabolic pathways. However, the analysis of protein abundance patterns associated with those pathways is complicated by the large number of proteins involved. In this study, the objective was to present the application of principal component analysis (PCA) to permit the visualization of developmental proteomic patterns of sarcoplasmic proteins found in breast muscle. Different turkey genotypes and nutritional regimens were used to potentially increase the variability within the sarcoplasmic protein profile. Sarcoplasmic protein fractions from turkey breast muscle samples were collected at 6 ages between 7 to 24 d. Breast muscle samples were collected from 2 distinctly different turkey lines. The poults within each line were either ad libitum or restrict fed. Proteomic PCA plots showed a visual developmental pattern from 7 until 17 d. Multivariate ANOVA highlighted the effect of time point and feeding regimen among profile patterns. The use of different genotypes and feeding regimens influenced variability, which was measured by mean Euclidean distances and ellipses of the PCA plots. These treatment effects, however, did not mask the developmental patterns. After 17 d, the proteomic patterns converged, suggesting that a level of biological stability was achieved regardless of the genotype or treatment. The developmental pattern obtained by the PCA methodology can aid in the planning of more efficient experimental designs so the developmental stage of individuals can be more accurately assessed.

Producing a one-dimensional proteomic map for human liver cytochromes p450

In this chapter we explore the inducible cytochrome P450 (CYP) forms as an example of membrane proteins analysis that relies on sodium dodecyl sulfate polyacrylamide gel electrophoresis (SDS-PAGE) fractionation with subsequent mass spectrometric (MS) identification. The approach involves cutting an SDS-PAGE gel lane into thin slices and identifying proteins in each slice by MS with the aim of obtaining detailed information on proteins of interest. A one-dimensional proteomic map showing the distribution of selected CYP isoforms across 40 slices was constructed using mass spectra obtained from each slice. Our protocol proved to be efficient enough to obtain a comprehensive profile of drug-metabolizing enzymes in the human liver. In addition to human tissues, the approach described should be applicable to the characterization of membrane proteins in other eukaryotic species.

Could LogP be a principal determinant of biological activity in 18-crown-6 ethers? Synthesis of biologically active adamantane-substituted diaza-crowns

18-crown-6 ethers are known to exert their biological activity by transporting K(+) ions across cell membranes. Using non-linear Support Vector Machines regression, we searched for structural features that influence antiproliferative activity in a diverse set of 19 known oxa-, monoaza- and diaza-18-crown-6 ethers. Here, we show that the logP of the molecule is the most important molecular descriptor, among ∼1300 tested descriptors, in determining biological potency (R(2)(cv) = 0.704). The optimal logP was at 5.5 (Ghose-Crippen ALOGP estimate) while both higher and lower values were detrimental to biological potency. After controlling for logP, we found that the antiproliferative activity of the molecule was generally not affected by side chain length, molecular symmetry, or presence of side chain amide links. To validate this QSAR model, we synthesized six novel, highly lipophilic diaza-18-crown-6 derivatives with adamantane moieties attached to the side arms. These compounds have near-optimal logP values and consequently exhibit strong growth inhibition in various human cancer cell lines and a bacterial system. The bioactivities of different diaza-18-crown-6 analogs in Bacillus subtilis and cancer cells were correlated, suggesting conserved molecular features may be mediating the cytotoxic response. We conclude that relying primarily on the logP is a sensible strategy in preparing future 18-crown-6 analogs with optimized biological activity.

Proteomics of plant pathogenic fungi

Plant pathogenic fungi cause important yield losses in crops. In order to develop efficient and environmental friendly crop protection strategies, molecular studies of the fungal biological cycle, virulence factors, and interaction with its host are necessary. For that reason, several approaches have been performed using both classical genetic, cell biology, and biochemistry and the modern, holistic, and high-throughput, omic techniques. This work briefly overviews the tools available for studying Plant Pathogenic Fungi and is amply focused on MS-based Proteomics analysis, based on original papers published up to December 2009. At a methodological level, different steps in a proteomic workflow experiment are discussed. Separate sections are devoted to fungal descriptive (intracellular, subcellular, extracellular) and differential expression proteomics and interactomics. From the work published we can conclude that Proteomics, in combination with other techniques, constitutes a powerful tool for providing important information about pathogenicity and virulence factors, thus opening up new possibilities for crop disease diagnosis and crop protection.

Application of slicing of one-dimensional gels with subsequent slice-by-slice mass spectrometry for the proteomic profiling of human liver cytochromes P450

Sequential thin slicing of one-dimensional electrophoresis gels followed by slice-by-slice mass spectrometry to allow protein identification was used to produce a proteomic map for cytochromes P450. Parallel MALDI-TOF-MS and LC-MS/MS analyses were performed. Combination of the two MS methods increased the quality of protein identification. We have proposed an efficient approach to obtain a comprehensive profile of drug-metabolizing enzymes in the liver that can be used to differentiate between polymorphic variants of cytochromes P450.