Pre-processing liquid chromatography/high-resolution mass spectrometry data: extracting pure mass spectra by deconvolution from the invariance of isotopic distribution.

RATIONALE: Mass spectra obtained by deconvolution of liquid chromatography/high-resolution mass spectrometry (LC/HRMS) data can be impaired by non-informative mass-over-charge (m/z) channels. This impairment of mass spectra can have significant negative influence on further post-processing, like quantification and identification. METHODS: A metric derived from the knowledge of errors in isotopic distribution patterns, and quality of the signal within a pre-defined mass chromatogram block, has been developed to pre-select all informative m/z channels. RESULTS: This procedure results in the clean-up of deconvoluted mass spectra by maintaining the intensity counts from m/z channels that originate from a specific compound/molecular ion, for example, molecular ion, adducts, (13) C-isotopes, multiply charged ions and removing all m/z channels that are not related to the specific peak. The methodology has been successfully demonstrated for two sets of high-resolution LC/MS data. CONCLUSIONS: The approach described is therefore thought to be a useful tool in the automatic processing of LC/HRMS data. It clearly shows the advantages compared to other approaches like peak picking and de-isotoping in the sense that all information is retained while non-informative data is removed automatically.

Instrument and process independent binning and baseline correction methods for liquid chromatography-high resolution-mass spectrometry deconvolution.

Setting appropriate bin sizes to aggregate hyphenated high-resolution mass spectrometry data, belonging to similar mass over charge (m/z) channels, is vital to metabolite quantification and further identification. In a high-resolution mass spectrometer when mass accuracy (ppm) varies as a function of molecular mass, which usually is the case while reading m/z from low to high values, it becomes a challenge to determine suitable bin sizes satisfying all m/z ranges. Similarly, the chromatographic process within a hyphenated system, like any other controlled processes, introduces some process driven systematic behavior that ultimately distorts the mass chromatogram signal. This is especially seen in liquid chromatogram-mass spectrometry (LC-MS) measurements where the gradient of the solvent and the washing step cycle-part of the chromatographic process, produce a mass chromatogram with a non-uniform baseline along the retention time axis. Hence prior to any automatic signal decomposition techniques like deconvolution, it is a equally vital to perform the baseline correction step for absolute metabolite quantification. This paper will discuss an instrument and process independent solution to the binning and the baseline correction problem discussed above, seen together, as an effective pre-processing step toward liquid chromatography-high resolution-mass spectrometry (LC-HR-MS) data deconvolution.

Evaluation of multiple variate selection methods from a biological perspective: a nutrigenomics case study.

Genomics-based technologies produce large amounts of data. To interpret the results and identify the most important variates related to phenotypes of interest, various multivariate regression and variate selection methods are used. Although inspected for statistical performance, the relevance of multivariate models in interpreting biological data sets often remains elusive. We compare various multivariate regression and variate selection methods applied to a nutrigenomics data set in terms of performance, utility and biological interpretability. The studied data set comprised hepatic transcriptome (10,072 predictor variates) and plasma protein concentrations [2 dependent variates: Leptin (LEP) and Tissue inhibitor of metalloproteinase 1 (TIMP-1)] collected during a high-fat diet study in ApoE3Leiden mice. The multivariate regression methods used were: partial least squares "PLS"; a genetic algorithm-based multiple linear regression, "GA-MLR"; two least-angle shrinkage methods, "LASSO" and "ELASTIC NET"; and a variant of PLS that uses covariance-based variate selection, "CovProc." Two methods of ranking the genes for Gene Set Enrichment Analysis (GSEA) were also investigated: either by their correlation with the protein data or by the stability of the PLS regression coefficients. The regression methods performed similarly, with CovProc and GA performing the best and worst, respectively (R-squared values based on "double cross-validation" predictions of 0.762 and 0.451 for LEP; and 0.701 and 0.482 for TIMP-1). CovProc, LASSO and ELASTIC NET all produced parsimonious regression models and consistently identified small subsets of variates, with high commonality between the methods. Comparison of the gene ranking approaches found a high degree of agreement, with PLS-based ranking finding fewer significant gene sets. We recommend the use of CovProc for variate selection, in tandem with univariate methods, and the use of correlation-based ranking for GSEA-like pathway analysis methods.

Comprehensive analysis of umami compounds by ion-pair liquid chromatography coupled to mass spectrometry.

UNLABELLED: An ion-pair LC-ESI-MS method was developed capable of analyzing various reported umami or umami-enhancing compounds, including glutamic acid and 5'-ribonucleotides. The method was validated using tomato and potato samples and showed overall good analytical performance with respect to selectivity, detection limit, linearity, and repeatability. The method was applied to various tomato samples resulting in concentrations of glutamic acid and 5'-ribonucleotides that were in good comparison with literature. The methodology might also be used for the discovery of new umami (enhancing) compounds in an untargeted mode. This was to a certain extent demonstrated for tomato samples by correlating all peaks observed with the ion-pair liquid chromatography-mass spectrometry (LC-MS) method to sensory properties using multivariate statistics. PRACTICAL APPLICATION: This study describes the development and application of a LC-MS method, which can be used to quantify several known umami (enhancing) compounds in various foods. Furthermore, the method might be useful for the discovery of new umami (enhancing) compounds.

Adipose tissue dysfunction signals progression of hepatic steatosis towards nonalcoholic steatohepatitis in C57BL/6 mice.

OBJECTIVE: Nonalcoholic fatty liver disease (NAFLD) is linked to obesity and diabetes, suggesting an important role of adipose tissue in the pathogenesis of NAFLD. Here, we aimed to investigate the interaction between adipose tissue and liver in NAFLD and identify potential early plasma markers that predict nonalcoholic steatohepatitis (NASH). RESEARCH DESIGN AND METHODS: C57Bl/6 mice were chronically fed a high-fat diet to induce NAFLD and compared with mice fed a low-fat diet. Extensive histological and phenotypical analyses coupled with a time course study of plasma proteins using multiplex assay were performed. RESULTS: Mice exhibited pronounced heterogeneity in liver histological scoring, leading to classification into four subgroups: low-fat low (LFL) responders displaying normal liver morphology, low-fat high (LFH) responders showing benign hepatic steatosis, high-fat low (HFL) responders displaying pre-NASH with macrovesicular lipid droplets, and high fat high (HFH) responders exhibiting overt NASH characterized by ballooning of hepatocytes, presence of Mallory bodies, and activated inflammatory cells. Compared with HFL responders, HFH mice gained weight more rapidly and exhibited adipose tissue dysfunction characterized by decreased final fat mass, enhanced macrophage infiltration and inflammation, and adipose tissue remodeling. Plasma haptoglobin, IL-1ß, TIMP-1, adiponectin, and leptin were significantly changed in HFH mice. Multivariate analysis indicated that in addition to leptin, plasma CRP, haptoglobin, eotaxin, and MIP-1α early in the intervention were positively associated with liver triglycerides. Intermediate prognostic markers of liver triglycerides included IL-18, IL-1ß, MIP-1γ, and MIP-2, whereas insulin, TIMP-1, granulocyte chemotactic protein 2, and myeloperoxidase emerged as late markers. CONCLUSIONS: Our data support the existence of a tight relationship between adipose tissue dysfunction and NASH pathogenesis and point to several novel potential predictive biomarkers for NASH.

The association of 83 plasma proteins with CHD mortality, BMI, HDL-, and total-cholesterol in men: applying multivariate statistics to identify proteins with prognostic value and biological relevance.

In this study, we applied the multivariate statistical tool Partial Least Squares (PLS) to analyze the relative importance of 83 plasma proteins in relation to coronary heart disease (CHD) mortality and the intermediate end points body mass index, HDL-cholesterol and total cholesterol. From a Dutch monitoring project for cardiovascular disease risk factors, men who died of CHD between initial participation (1987-1991) and end of follow-up (January 1, 2000) (N = 44) and matched controls (N = 44) were selected. Baseline plasma concentrations of proteins were measured by a multiplex immunoassay. With the use of PLS, we identified 15 proteins with prognostic value for CHD mortality and sets of proteins associated with the intermediate end points. Subsequently, sets of proteins and intermediate end points were analyzed together by Principal Components Analysis, indicating that proteins involved in inflammation explained most of the variance, followed by proteins involved in metabolism and proteins associated with total-C. This study is one of the first in which the association of a large number of plasma proteins with CHD mortality and intermediate end points is investigated by applying multivariate statistics, providing insight in the relationships among proteins, intermediate end points and CHD mortality, and a set of proteins with prognostic value.

Improving the analysis of designed studies by combining statistical modelling with study design information.

BACKGROUND: In the fields of life sciences, so-called designed studies are used for studying complex biological systems. The data derived from these studies comply with a study design aimed at generating relevant information while diminishing unwanted variation (noise). Knowledge about the study design can be used to decompose the total data into data blocks that are associated with specific effects. Subsequent statistical analysis can be improved by this decomposition if these are applied on selected combinations of effects. RESULTS: The benefit of this approach was demonstrated with an analysis that combines multivariate PLS (Partial Least Squares) regression with data decomposition from ANOVA (Analysis of Variance): ANOVA-PLS. As a case, a nutritional intervention study is used on Apoliprotein E3-Leiden (APOE3Leiden) transgenic mice to study the relation between liver lipidomics and a plasma inflammation marker, Serum Amyloid A. The ANOVA-PLS performance was compared to PLS regression on the non-decomposed data with respect to the quality of the modelled relation, model reliability, and interpretability. CONCLUSION: It was shown that ANOVA-PLS leads to a better statistical model that is more reliable and better interpretable compared to standard PLS analysis. From a following biological interpretation, more relevant metabolites were derived from the model. The concept of combining data composition with a subsequent statistical analysis, as in ANOVA-PLS, is however not limited to PLS regression in metabolomics but can be applied for many statistical methods and many different types of data.

Transcriptomics analysis of interactive effects of benzene, trichloroethylene and methyl mercury within binary and ternary mixtures on the liver and kidney following subchronic exposure in the rat.

The present research aimed to study the interaction of three chemicals, methyl mercury, benzene and trichloroethylene, on mRNA expression alterations in rat liver and kidney measured by microarray analysis. These compounds were selected based on presumed different modes of action. The chemicals were administered daily for 14 days at the Lowest-Observed-Adverse-Effect-Level (LOAEL) or at a two- or threefold lower concentration individually or in binary or ternary mixtures. The compounds had strong antagonistic effects on each other's gene expression changes, which included several genes encoding Phase I and II metabolizing enzymes. On the other hand, the mixtures affected the expression of "novel" genes that were not or little affected by the individual compounds. The three compounds exhibited a synergistic interaction on gene expression changes at the LOAEL in the liver and both at the sub-LOAEL and LOAEL in the kidney. Many of the genes induced by mixtures but not by single compounds, such as Id2, Nr2f6, Tnfrsf1a, Ccng1, Mdm2 and Nfkb1 in the liver, are known to affect cellular proliferation, apoptosis and tissue-specific function. This indicates a shift from compound specific response on exposure to individual compounds to a more generic stress response to mixtures. Most of the effects on cell viability as concluded from transcriptomics were not detected by classical toxicological endpoints illustrating the benefit of increased sensitivity of assessing gene expression profiling. These results emphasize the benefit of applying toxicogenomics in mixture interaction studies, which yields biomarkers for joint toxicity and eventually can result in an interaction model for most known toxicants.

Effects of growth conditions and processing on Rehmannia glutinosa using fingerprint strategy.

Metabolite profiling in combination with multivariate statistics is a sophisticated method for quality assessment of natural products. For the development of a quality control strategy in Traditional Chinese Medicine (TCM), we have measured the metabolite fingerprints of Rehmannia glutinosa by GC-MS. Plants were grown under different climate and soil conditions in a phytotron and were processed by a variable number of repetitive steps to investigate the effects on both growth conditions and processing for material medica of R. glutinosa. The GC-MS data have been analyzed by principal component analysis (PCA) and the new approach of the ANOVA-simultaneous component analysis (ASCA) which can combine the information from a structured data design with multivariate analysis. The results clearly show the effect of the different factors and indicate directions for process improvement. When plants were grown under various temperatures, humidity and light intensities for a short period (3 weeks), no significant changes on studied metabolites were observed. However, significant changes were found between different processing cycles. The present data clearly indicate the importance of strictly controlling processing in R. glutinosa and illustrate the impact of growth conditions. This is the first report on the metabolite profile of R. glutinosa that provides a base for the establishment of a quality control strategy.

Multivariate calibration with least-squares support vector machines.

This paper proposes the use of least-squares support vector machines (LS-SVMs) as a relatively new nonlinear multivariate calibration method, capable of dealing with ill-posed problems. LS-SVMs are an extension of "traditional" SVMs that have been introduced recently in the field of chemistry and chemometrics. The advantages of SVM-based methods over many other methods are that these lead to global models that are often unique, and nonlinear regression can be performed easily as an extension to linear regression. An additional advantage of LS-SVM (compared to SVM) is that model calculation and optimization can be performed relatively fast. As a test case to study the use of LS-SVM, the well-known and important chemical problem is considered in which spectra are affected by nonlinear interferences. As one specific example, a commonly used case is studied in which near-infrared spectra are affected by temperature-induced spectral variation. Using this test case, model optimization, pruning, and model interpretation of the LS-SVM have been demonstrated. Furthermore, excellent performance of the LS-SVM, compared to other approaches, has been presented on the specific example. Therefore, it can be concluded that LS-SVMs can be seen as very promising techniques to solve ill-posed problems. Furthermore, these have been shown to lead to robust models in cases of spectral variations due to nonlinear interferences.