Multiple comparisons in mass-spectrometry-based -omics technologies

Comparison of chemometric strategies for potential exposure marker discovery and false-positive reduction in untargeted metabolomics: application to the serum analysis by LC-HRMS after intake of Vaccinium fruit supplements

Untargeted liquid chromatographic-high-resolution mass spectrometric (LC-HRMS) metabolomics for potential exposure marker (PEM) discovery in nutrikinetic studies generates complex outputs. The correct selection of statistically significant PEMs is a crucial analytical step for understanding nutrition-health interactions. Hence, in this paper, different chemometric selection workflows for PEM discovery, using multivariate or univariate parametric or non-parametric data analyses, were comparatively tested and evaluated. The PEM selection protocols were applied to a small-sample-size untargeted LC-HRMS study of a longitudinal set of serum samples from 20 volunteers after a single intake of (poly)phenolic-rich Vaccinium myrtillus and Vaccinium corymbosum supplements. The non-parametric Games-Howell test identified a restricted group of significant features, thus minimizing the risk of false-positive retention. Among the forty-seven PEMs exhibiting a statistically significant postprandial kinetics, twelve were successfully annotated as purine pathway metabolites, benzoic and benzodiol metabolites, indole alkaloids, and organic and fatty acids, and five (i.e. octahydro-methyl-β-carboline-dicarboxylic acid, tetrahydro-methyl-β-carboline-dicarboxylic acid, citric acid, caprylic acid, and azelaic acid) were associated to Vaccinium berry consumption for the first time. The analysis of the area under the curve of the longitudinal dataset highlighted thirteen statistically significant PEMs discriminating the two interventions, including four intra-intervention relevant metabolites (i.e. abscisic acid glucuronide, catechol sulphate, methyl-catechol sulphate, and α-hydroxy-hippuric acid). Principal component analysis and sample classification through linear discriminant analysis performed on PEM maximum intensity confirmed the discriminating role of these PEMs.

MultipleTesting.com: A tool for life science researchers for multiple hypothesis testing correction

Scientists from nearly all disciplines face the problem of simultaneously evaluating many hypotheses. Conducting multiple comparisons increases the likelihood that a non-negligible proportion of associations will be false positives, clouding real discoveries. Drawing valid conclusions require taking into account the number of performed statistical tests and adjusting the statistical confidence measures. Several strategies exist to overcome the problem of multiple hypothesis testing. We aim to summarize critical statistical concepts and widely used correction approaches while also draw attention to frequently misinterpreted notions of statistical inference. We provide a step-by-step description of each multiple-testing correction method with clear examples and present an easy-to-follow guide for selecting the most suitable correction technique. To facilitate multiple-testing corrections, we developed a fully automated solution not requiring programming skills or the use of a command line. Our registration free online tool is available at www.multipletesting.com and compiles the five most frequently used adjustment tools, including the Bonferroni, the Holm (step-down), the Hochberg (step-up) corrections, allows to calculate False Discovery Rates (FDR) and q-values. The current summary provides a much needed practical synthesis of basic statistical concepts regarding multiple hypothesis testing in a comprehensible language with well-illustrated examples. The web tool will fill the gap for life science researchers by providing a user-friendly substitute for command-line alternatives.

A methodological approach to correlate tumor heterogeneity with drug distribution profile in mass spectrometry imaging data

BACKGROUND

Drug mass spectrometry imaging (MSI) data contain knowledge about drug and several other molecular ions present in a biological sample. However, a proper approach to fully explore the potential of such type of data is still missing. Therefore, a computational pipeline that combines different spatial and non-spatial methods is proposed to link the observed drug distribution profile with tumor heterogeneity in solid tumor. Our data analysis steps include pre-processing of MSI data, cluster analysis, drug local indicators of spatial association (LISA) map, and ions selection.

RESULTS

The number of clusters identified from different tumor tissues. The spatial homogeneity of the individual cluster was measured using a modified version of our drug homogeneity method. The clustered image and drug LISA map were simultaneously analyzed to link identified clusters with observed drug distribution profile. Finally, ions selection was performed using the spatially aware method.

CONCLUSIONS

In this paper, we have shown an approach to correlate the drug distribution with spatial heterogeneity in untargeted MSI data. Our approach is freely available in an R package 'CorrDrugTumorMSI'.

Cyto-Feature Engineering: A Pipeline for Flow Cytometry Analysis to Uncover Immune Populations and Associations with Disease

Flow cytometers can now analyze up to 50 parameters per cell and millions of cells per sample; however, conventional methods to analyze data are subjective and time-consuming. To address these issues, we have developed a novel flow cytometry analysis pipeline to identify a plethora of cell populations efficiently. Coupled with feature engineering and immunological context, researchers can immediately extrapolate novel discoveries through easy-to-understand plots. The R-based pipeline uses Fluorescence Minus One (FMO) controls or distinct population differences to develop thresholds for positive/negative marker expression. The continuous data is transformed into binary data, capturing a positive/negative biological dichotomy often of interest in characterizing cells. Next, a filtering step refines the data from all identified cell phenotypes to populations of interest. The data can be partitioned by immune lineages and statistically correlated to other experimental measurements. The pipeline’s modularity allows customization of statistical testing, adoption of alternative initial gating steps, and incorporation of other datasets. Validation of this pipeline through manual gating of two datasets (murine splenocytes and human whole blood) confirmed its accuracy in identifying even rare subsets. Lastly, this pipeline can be applied in all disciplines utilizing flow cytometry regardless of cytometer or panel design. The code is available at https://github.com/aef1004/cyto-feature_engineering.

Assessing the Similarity between Random Copolymer Drug Glatiramer Acetate by Using LC-MS Data Coupling with Hypothesis Testing

The full characterization of nonbiological complex drugs (NBCDs) is not possible, but analytical approaches are of urgent need to evaluate the similarity between different lots and compare with their follow-up versions. Here, we propose a hypothesis testing-based approach to assess the similarity/difference between random amino acid copolymer drugs using liquid chromatography mass spectrometry (LC-MS) analysis. Two glatiramer acetate (GA) drugs, commercially available Copaxone and in-house synthesized SPT, and a negative control were digested by Lys-C and followed by HILIC-MS analysis. After retention time alignment and feature identification, 1627 features matched to m/z values in an elemental composition database were considered as derived from active drug ingredients. A hypothesis testing approach, the sum of squared deviations test, was developed to process high-dimensional data derived from LC-MS spectra. The feasibility of this approach was first demonstrated by testing 5 versus 5 lots of Copaxone and Copaxone versus SPT, which suggested a significant similarity by obtaining the estimated 95th percentile of the distribution of the estimator (ρ̂_(95%)) at 0.0056 (p-value = 0.0026) and 0.0026 (p-value < 0.0001), respectively. In contrast, the ρ̂ was 0.036 (p-value = 1.00), while comparing Copaxone and the negative control, implying a lack of similarity. We further synthesized nine stable isotope-labeled peptides to validate the proposed amino acid sequences in the database, demonstrating the correctness in sequence identification. The quantitation variations in our analytical procedures were determined to be 6.8-7.7%. This approach was found to have a great potential for evaluating the similarity between generic NBCDs and listed reference drugs, as well as to monitor the lot-to-lot variation.

Gender specific decrease of a set of circulating N-acylphosphatidyl ethanolamines (NAPEs) in the plasma of Parkinson's disease patients

INTRODUCTION

Current markers of Parkinson's disease (PD) fail to detect the early progression of disease state. Conversely, current omics techniques allow the investigation of hundreds of molecules potentially altered by disease conditions. Based on evidence previously collected by our group in a mouse model of PD, we speculated that a particular set of circulating lipids might be significantly altered by the pathology.

OBJECTIVES

The aim of current study was to evaluate the potential of a particular set of N-acyl-phosphatidylethanolamines (NAPEs) as potential non-invasive plasma markers of ongoing neurodegeneration from Parkinson's disease in human subjects.

METHODS

A panel of seven NAPEs were quantified by LC-MS/MS in the plasma of 587 individuals (healthy controls, n = 319; Parkinson's disease, n = 268); Random Forest classification and statistical modeling was applied to compare Parkinson's disease versus controls. All p-values obtained in different tests were corrected for multiplicity by controlling the false discovery rate (FDR).

RESULTS

The results indicate that this panel of NAPEs is able to distinguish female PD patients from the corresponding healthy controls. Further to this, the observed downregulation of these NAPEs is in line with the results in plasma of a mouse model of Parkinson's (6-OHDA).

CONCLUSIONS

In the current study we have shown the downregulation of NAPEs in plasma of PD patients and we thus speculate that these lipids might serve as candidate biomarkers for PD. We also suggest a molecular mechanism, explaining our findings, which involves gut microbiota.

A practical tool for maximal information coefficient analysis

Background

The ability of finding complex associations in large omics datasets, assessing their significance, and prioritizing them according to their strength can be of great help in the data exploration phase. Mutual information-based measures of association are particularly promising, in particular after the recent introduction of the TIC_e and MIC_e estimators, which combine computational efficiency with superior bias/variance properties. An open-source software implementation of these two measures providing a complete procedure to test their significance would be extremely useful.

Findings

Here, we present MICtools, a comprehensive and effective pipeline that combines TIC_e and MIC_e into a multistep procedure that allows the identification of relationships of various degrees of complexity. MICtools calculates their strength assessing statistical significance using a permutation-based strategy. The performances of the proposed approach are assessed by an extensive investigation in synthetic datasets and an example of a potential application on a metagenomic dataset is also illustrated.

Conclusions

We show that MICtools, combining TIC_e and MIC_e, is able to highlight associations that would not be captured by conventional strategies.

Nutrimetabolomics: An Integrative Action for Metabolomic Analyses in Human Nutritional Studies

The life sciences are currently being transformed by an unprecedented wave of developments in molecular analysis, which include important advances in instrumental analysis as well as biocomputing. In light of the central role played by metabolism in nutrition, metabolomics is rapidly being established as a key analytical tool in human nutritional studies. Consequently, an increasing number of nutritionists integrate metabolomics into their study designs. Within this dynamic landscape, the potential of nutritional metabolomics (nutrimetabolomics) to be translated into a science, which can impact on health policies, still needs to be realized. A key element to reach this goal is the ability of the research community to join, to collectively make the best use of the potential offered by nutritional metabolomics. This article, therefore, provides a methodological description of nutritional metabolomics that reflects on the state-of-the-art techniques used in the laboratories of the Food Biomarker Alliance (funded by the European Joint Programming Initiative "A Healthy Diet for a Healthy Life" (JPI HDHL)) as well as points of reflections to harmonize this field. It is not intended to be exhaustive but rather to present a pragmatic guidance on metabolomic methodologies, providing readers with useful "tips and tricks" along the analytical workflow.

Statistical and Multivariate Analysis of MS-Based Plant Metabolomics Data

Raw data from metabolomics experiments are initially subjected to peak identification and signal deconvolution to generate raw data matrices m × n, where m are samples and n are metabolites. We describe here simple statistical procedures on such multivariate data matrices, all provided as functions in the programming environment R, useful to normalize data, detect biomarkers, and perform sample classification.

Applying generalized additive models to unravel dynamic changes in anthocyanin biosynthesis in methyl jasmonate elicited grapevine (Vitis vinifera cv. Gamay) cell cultures

Plant cell cultures represent important model systems to understand metabolism and its modulation by regulatory factors. Even in controlled conditions, cell metabolism is highly dynamic and can be fully characterized only by time course experiments. Here, we show that statistical analysis of this type of data gains power if it moves to approaches able to compare the 'trends' of the different metabolites. In particular, we show how generalized additive models can be used to model the time-dependent profile of anthocyanin synthesis in grapevine cell suspension cultures (Vitis vinifera cv. Gamay), following treatment with 100 μm methyl jasmonate. The sampling was performed daily for 20 days of culturing following elicitation at day 5. All samples were analyzed by UPLC-MS/MS for the identification and quantification of fifteen anthocyanin compounds. The models confirmed the separation in the anthocyanin biosynthetic pathway between delphinidin-based and cyanidin-based compounds, showing that methyl jasmonate modulates the anthocyanin concentration profiles. Our results clearly indicate that the combination of high-throughput metabolomics and state of the art statistical modeling is a powerful approach to study plant metabolism. This approach is expected to gain popularity due to the growing availability of low-cost high-throughput 'omic' assays.

Integration of Traditional and Metabolomics Biomarkers Identifies Prognostic Metabolites for Predicting Responsiveness to Nutritional Intervention against Oxidative Stress and Inflammation

Various statistical approaches can be applied to integrate traditional and omics biomarkers, allowing the discovery of prognostic markers to classify subjects into poor and good prognosis groups in terms of responses to nutritional interventions. Here, we performed a prototype study to identify metabolites that predict responses to an intervention against oxidative stress and inflammation, using a data set from a randomized controlled trial evaluating Korean black raspberry (KBR) in sedentary overweight/obese subjects. First, a linear mixed-effects model analysis with multiple testing correction showed that four-week consumption of KBR significantly changed oxidized glutathione (GSSG, q = 0.027) level, the ratio of reduced glutathione (GSH) to GSSG (q = 0.039) in erythrocytes, malondialdehyde (MDA, q = 0.006) and interleukin-6 (q = 0.006) levels in plasma, and seventeen NMR metabolites in urine compared with those in the placebo group. A subsequent generalized linear mixed model analysis showed linear correlations between baseline urinary glycine and N-phenylacetylglycine (PAG) and changes in the GSH:GSSG ratio (p = 0.008 and 0.004) as well as between baseline urinary adenine and changes in MDA (p = 0.018). Then, receiver operating characteristic analysis revealed that a two-metabolite set (glycine and PAG) had the strongest prognostic relevance for future interventions against oxidative stress (the area under the curve (AUC) = 0.778). Leave-one-out cross-validation confirmed the accuracy of prediction (AUC = 0.683). The current findings suggest that a higher level of this two-metabolite set at baseline is useful for predicting responders to dietary interventions in subjects with oxidative stress and inflammation, contributing to the emergence of personalized nutrition.

Regional features of northern Italian sparkling wines, identified using solid-phase micro extraction and comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry

We carried out comprehensive mapping of volatile compounds in 70 wines, from 48 wineries and 6 vintages, representative of the two main production areas for Italian sparkling wines, by HS-SPME-GCxGC-TOF-MS and multivariate analysis. The final scope was to describe the metabolomics space of these wines, and to verify whether the grape cultivar signature, the pedoclimatic influence of the production area, and the complex technology were measurable in the final product. The wine chromatograms provided a wealth of information, with 1695 compounds being found. A large number of putative markers influenced by the cultivation area was observed. A subset of 196 biomarkers fully discriminated between the two types of sparkling wines investigated. Among the new compounds, safranal and α-isophorone were observed. We showed how correlation-based network analysis could be used as a tool to detect the differences in compound behaviour based on external/environmental influences.

Comparative lipidomic study of urothelial cancer models: association with urothelial cancer cell invasiveness

A joint NMR/LC-MS approach allows to establish significant differences in the lipidoma of invasive urothelial carcinoma cells (T24) with respect to noninvasive urothelial cells (RT4).

MetaDB a Data Processing Workflow in Untargeted MS-Based Metabolomics Experiments

Due to their sensitivity and speed, mass-spectrometry based analytical technologies are widely used to in metabolomics to characterize biological phenomena. To address issues like metadata organization, quality assessment, data processing, data storage, and, finally, submission to public repositories, bioinformatic pipelines of a non-interactive nature are often employed, complementing the interactive software used for initial inspection and visualization of the data. These pipelines often are created as open-source software allowing the complete and exhaustive documentation of each step, ensuring the reproducibility of the analysis of extensive and often expensive experiments. In this paper, we will review the major steps which constitute such a data processing pipeline, discussing them in the context of an open-source software for untargeted MS-based metabolomics experiments recently developed at our institute. The software has been developed by integrating our metaMS R package with a user-friendly web-based application written in Grails. MetaMS takes care of data pre-processing and annotation, while the interface deals with the creation of the sample lists, the organization of the data storage, and the generation of survey plots for quality assessment. Experimental and biological metadata are stored in the ISA-Tab format making the proposed pipeline fully integrated with the Metabolights framework.