Data processing is one of the biggest problems in metabolomics, given the high number of samples analyzed and the need of multiple software packages for each step of the processing workflow.
Objectives
Merge in the same platform the steps required for metabolomics data processing.
Methods
KniMet is a workflow for the processing of mass spectrometry-metabolomics data based on the KNIME Analytics platform.
Results
The approach includes key steps to follow in metabolomics data processing: feature filtering, missing value imputation, normalization, batch correction and annotation.
Conclusion
KniMet provides the user with a local, modular and customizable workflow for the processing of both GC–MS and LC–MS open profiling data.
Untargeted metabolomics studies for biomarker discovery often have hundreds to thousands of human samples. Data acquisition of large-scale samples has to be divided into several batches and may span from months to as long as several years. The signal drift of metabolites during data acquisition (intra- and inter-batch) is unavoidable and is a major confounding factor for large-scale metabolomics studies.
Objectives
We aim to develop a data normalization method to reduce unwanted variations and integrate multiple batches in large-scale metabolomics studies prior to statistical analyses.
Methods
We developed a machine learning algorithm-based method, support vector regression (SVR), for large-scale metabolomics data normalization and integration. An R package named MetNormalizer was developed and provided for data processing using SVR normalization.
Results
After SVR normalization, the portion of metabolite ion peaks with relative standard deviations (RSDs) less than 30 % increased to more than 90 % of the total peaks, which is much better than other common normalization methods. The reduction of unwanted analytical variations helps to improve the performance of multivariate statistical analyses, both unsupervised and supervised, in terms of classification and prediction accuracy so that subtle metabolic changes in epidemiological studies can be detected.
Conclusion
SVR normalization can effectively remove the unwanted intra- and inter-batch variations, and is much better than other common normalization methods.
Data sharing is being increasingly required by journals and has been heralded as a solution to the ‘replication crisis’.
Objectives
(i) Review data sharing policies of journals publishing the most metabolomics papers associated with open data and (ii) compare these journals’ policies to those that publish the most metabolomics papers.
Methods
A PubMed search was used to identify metabolomics papers. Metabolomics data repositories were manually searched for linked publications.
Results
Journals that support data sharing are not necessarily those with the most papers associated to open metabolomics data.
Conclusion
Further efforts are required to improve data sharing in metabolomics.
Although cultured cells are nowadays regularly analyzed by metabolomics technologies, some issues in study setup and data processing are still not resolved to complete satisfaction: a suitable harvesting method for adherent cells, a fast and robust method for data normalization, and the proof that metabolite levels can be normalized to cell number.
Objectives
We intended to develop a fast method for normalization of cell culture metabolomics samples, to analyze how metabolite levels correlate with cell numbers, and to elucidate the impact of the kind of harvesting on measured metabolite profiles.
Methods
We cultured four different human cell lines and used them to develop a fluorescence-based method for DNA quantification. Further, we assessed the correlation between metabolite levels and cell numbers and focused on the impact of the harvesting method (scraping or trypsinization) on the metabolite profile.
Results
We developed a fast, sensitive and robust fluorescence-based method for DNA quantification showing excellent linear correlation between fluorescence intensities and cell numbers for all cell lines. Furthermore, 82–97 % of the measured intracellular metabolites displayed linear correlation between metabolite concentrations and cell numbers. We observed differences in amino acids, biogenic amines, and lipid levels between trypsinized and scraped cells.
Conclusion
We offer a fast, robust, and validated normalization method for cell culture metabolomics samples and demonstrate the eligibility of the normalization of metabolomics data to the cell number. We show a cell line and metabolite-specific impact of the harvesting method on metabolite concentrations.
Urine is an ideal matrix for metabolomics investigation due to its non-invasive nature of collection and its rich metabolite content. Despite the advancements in mass spectrometry and 1H-NMR platforms in urine metabolomics, the statistical analysis of the generated data is challenged with the need to adjust for the hydration status of the person. Normalization to creatinine or osmolality values are the most adopted strategies, however, each technique has its challenges that can hinder its wider application. We have been developing targeted urine metabolomic methods to differentiate two important respiratory diseases, namely asthma and chronic obstructive pulmonary disease (COPD).
Objective
To assess whether the statistical model of separation of diseases using targeted metabolomic data would be improved by normalization to osmolality instead of creatinine.
Methods
The concentration of 32 metabolites was previously measured by two liquid chromatography-tandem mass spectrometry methods in 51 human urine samples with either asthma (n?=?25) or COPD (n?=?26). The data was normalized to creatinine or osmolality. Statistical analysis of the normalized values in each disease was performed using partial least square discriminant analysis (PLS-DA). Models of separation of diseases were compared.
Results
We found that normalization to creatinine or osmolality did not significantly change the PLS-DA models of separation (R2Q2?=?0.919, 0.705 vs R2Q2?=?0.929, 0.671, respectively). The metabolites of importance in the models remained similar for both normalization methods.
Conclusion
Our findings suggest that targeted urine metabolomic data can be normalized for hydration using creatinine or osmolality with no significant impact on the diagnostic accuracy of the model.
Untargeted metabolomics is a powerful tool for biological discoveries. To analyze the complex raw data, significant advances in computational approaches have been made, yet it is not clear how exhaustive and reliable the data analysis results are.
Objectives
Assessment of the quality of raw data processing in untargeted metabolomics.
Methods
Five published untargeted metabolomics studies, were reanalyzed.
Results
Omissions of at least 50 relevant compounds from the original results as well as examples of representative mistakes were reported for each study.
Conclusion
Incomplete raw data processing shows unexplored potential of current and legacy data.
Metabolomics is a well-established tool in systems biology, especially in the top–down approach. Metabolomics experiments often results in discovery studies that provide intriguing biological hypotheses but rarely offer mechanistic explanation of such findings. In this light, the interpretation of metabolomics data can be boosted by deploying systems biology approaches.
Objectives
This review aims to provide an overview of systems biology approaches that are relevant to metabolomics and to discuss some successful applications of these methods.
Methods
We review the most recent applications of systems biology tools in the field of metabolomics, such as network inference and analysis, metabolic modelling and pathways analysis.
Results
We offer an ample overview of systems biology tools that can be applied to address metabolomics problems. The characteristics and application results of these tools are discussed also in a comparative manner.
Conclusions
Systems biology-enhanced analysis of metabolomics data can provide insights into the molecular mechanisms originating the observed metabolic profiles and enhance the scientific impact of metabolomics studies.
In plant metabolomics, metabolite contents are often normalized by sample weight. However, accurate weighing of very small samples, such as individual Arabidopsis thaliana seeds (approximately 20 µg), is difficult, which may lead to irreproducible results.
Objectives
We aimed to establish alternative normalization methods for seed-grain-based comparative metabolomics of A. thaliana.
Methods
Arabidopsis thaliana seeds were assumed to have a prolate spheroid shape. Using a microscope image of each seed, the lengths of major and minor axes were measured by fitting a projected 2-dimensional shape of each seed as an ellipse. Metabolic profiles of individual diploid or tetraploid A. thaliana seeds were measured by our highly sensitive protocol (“widely targeted metabolomics”) that uses liquid chromatography coupled with tandem quadrupole mass spectrometry. Mass spectrometric analysis of 1 µL of solution extract identified more than 100 metabolites. The data were normalized by various seed-size measures, including seed volume (single-grain-based analysis). For comparison, metabolites were extracted from 4 mg of diploid and tetraploid A. thaliana seeds and their metabolic profiles were analyzed by normalization of weight (weight-based analysis).
Results
A small number of metabolites showed statistically significant differences in the single-grain-based analysis compared to weight-based analysis. A total of 17 metabolites showed statistically different accumulation between ploidy types with similar fold changes in both analyses.
Conclusion
Seed-size measures obtained by microscopic imaging were useful for data normalization. Single-grain-based analysis enables evaluation of metabolism of each seed and elucidates the metabolic profiles of precious bioresources by using small amounts of samples.
In metabolomics studies, unwanted variation inevitably arises from various sources. Normalization, that is the removal of unwanted variation, is an essential step in the statistical analysis of metabolomics data. However, metabolomics normalization is often considered an imprecise science due to the diverse sources of variation and the availability of a number of alternative strategies that may be implemented.
Objectives
We highlight the need for comparative evaluation of different normalization methods and present software strategies to help ease this task for both data-oriented and biological researchers.
Methods
We present NormalizeMets—a joint graphical user interface within the familiar Microsoft Excel and freely-available R software for comparative evaluation of different normalization methods. The NormalizeMets R package along with the vignette describing the workflow can be downloaded from https://cran.r-project.org/web/packages/NormalizeMets/. The Excel Interface and the Excel user guide are available on https://metabolomicstats.github.io/ExNormalizeMets.
Results
NormalizeMets allows for comparative evaluation of normalization methods using criteria that depend on the given dataset and the ultimate research question. Hence it guides researchers to assess, select and implement a suitable normalization method using either the familiar Microsoft Excel and/or freely-available R software. In addition, the package can be used for visualisation of metabolomics data using interactive graphical displays and to obtain end statistical results for clustering, classification, biomarker identification adjusting for confounding variables, and correlation analysis.
Conclusion
NormalizeMets is designed for comparative evaluation of normalization methods, and can also be used to obtain end statistical results. The use of freely-available R software offers an attractive proposition for programming-oriented researchers, and the Excel interface offers a familiar alternative to most biological researchers. The package handles the data locally in the user’s own computer allowing for reproducible code to be stored locally.
Failure to properly account for normal systematic variations in OMICS datasets may result in misleading biological conclusions. Accordingly, normalization is a necessary step in the proper preprocessing of OMICS datasets. In this regards, an optimal normalization method will effectively reduce unwanted biases and increase the accuracy of downstream quantitative analyses. But, it is currently unclear which normalization method is best since each algorithm addresses systematic noise in different ways.
Objective
Determine an optimal choice of a normalization method for the preprocessing of metabolomics datasets.
Methods
Nine MVAPACK normalization algorithms were compared with simulated and experimental NMR spectra modified with added Gaussian noise and random dilution factors. Methods were evaluated based on an ability to recover the intensities of the true spectral peaks and the reproducibility of true classifying features from orthogonal projections to latent structures—discriminant analysis model (OPLS-DA).
Results
Most normalization methods (except histogram matching) performed equally well at modest levels of signal variance. Only probabilistic quotient (PQ) and constant sum (CS) maintained the highest level of peak recovery (>?67%) and correlation with true loadings (>?0.6) at maximal noise.
Conclusion
PQ and CS performed the best at recovering peak intensities and reproducing the true classifying features for an OPLS-DA model regardless of spectral noise level. Our findings suggest that performance is largely determined by the level of noise in the dataset, while the effect of dilution factors was negligible. A minimal allowable noise level of 20% was also identified for a valid NMR metabolomics dataset.
Until recently, plant metabolomics have provided a deep understanding on the metabolic regulation in individual plants as experimental units. The application of these techniques to agricultural systems subjected to more complex interactions is a step towards the implementation of translational metabolomics in crop breeding.
Aim of Review
We present here a review paper discussing advances in the knowledge reached in the last years derived from the application of metabolomic techniques that evolved from biomarker discovery to improve crop yield and quality.
Key Scientific Concepts of Review
Translational metabolomics applied to crop breeding programs.
Recurrent Clostridium difficile infection (CDI) is associated with intestinal dysbiosis. Currently, there is no diagnostic test to predict at-risk patients for CDI recurrences. Urine metabolomics may have prognostic value, but have not been characterized in this patient population.
Objective
The aim of this pilot study was to profile the urine metabolomics of patients with various frequencies of CDI.
Methods
Spot urine samples were prospectively collected from 31 adults who at various stages of recurrent CDI (1 to >5 episodes). Patients were age- and sex-matched in a 1:1 ratio with healthy controls. Urine metabolomics was performed and spectra were assessed using Chenomx NMRSuite v7 and analyzed using multivariate statistics with MetaboAnalyst 3.0. Stool metagenomic analyses were performed in six patients with >3 episodes of CDI and compared to 7 healthy controls, which were correlated with urine metabolomics.
Results
Using 53 metabolites, a two-component, partial least squares—discriminant analysis (PLS-DA) was built that clearly discriminated between healthy controls and CDI patients. The anticipated gender-based difference was not found within the CDI patient group. However, separations between (1) healthy control and CDI patients, as well as (2) patients with different episodes of CDI were possible and the permutations found were significant. Furthermore, choline was found to be the single most important urine metabolite separating healthy controls from CDI patients, and the microbiota from recurrent CDI patients was found to have decreased abundance of choline metabolizing bacteria.
Conclusions
Using small groups in a preliminary study, we have demonstrated that urine metabolomics has the potential to distinguish between healthy controls and patients with CDI. Furthermore, it could discriminate between patients experiencing different frequencies of recurrent CDI. If validated in larger cohorts, urine metabolomics has potential at identifying patients who are at risk for recurrent CDI. The significance of choline-deficient microbiota in CDI patients should be further examined.
Recent advances in microbiome research have revealed the diverse participation of gut microbiota in a number of diseases. Bacteria-specific endogenous small molecules are produced in the gut, are transported throughout the whole body by circulation, and play key roles in disease establishment. However, the factors and mechanisms underlying these microbial influences largely remain unknown.
Objectives
The purpose of this study was to use metabolomics to better understand the influence of microbiota on host physiology.
Methods
Germ-free mice (GF) were orally administered with the feces of specific pathogen-free (SPF) mice and were maintained in a vinyl isolator for 4 weeks for establishing the so-called ExGF mice. Comparative metabolomics was performed on luminal contents, feces, urine, plasma, and tissues of GF and ExGF mice.
Results
The metabolomics profile of 1716 compounds showed marked difference between GF and ExGF for each matrix. Intestinal differences clearly showed the contribution of microbiota to host digestive activities. In addition, colonic metabolomics revealed the efficient conversion of primary to secondary metabolites by microbiota. Furthermore, metabolomics of tissues and excrements demonstrated the effect of microbiota on the accumulation of metabolites in tissues and during excretion. These effects included known bacterial effects (such as bile acids and amino acids) as well as novel ones, including a drastic decrease of sphingolipids in the host.
Conclusion
The diverse effects of microbiota on different sites of the host metabolome were revealed and novel influences on host physiology were demonstrated. These findings should contribute to a deeper understanding of the influence of gut microbiota on disease states and aid in the development of effective intervention strategies.
The clinical management of Gestational diabetes mellitus (GDM) would benefit from enhanced metabolic knowledge both at the time of diagnosis and during therapy.
Objectives
This work aimed at unveiling metabolic markers of GDM and of the subjects’ response to therapy.
Methods
Urine NMR metabolomics was used with a variable selection methodology to reduce uninformative variability. The NMR data was analysed by multivariate and univariate analysis methodologies.
Results
The results showed that urine NMR metabolomics enables a metabolic signature of GDM to be identified at the time of diagnosis. This signature comprises relevant changes in 12 NMR metabolites/resonances and qualitative variations in a number of additional metabolites. The metabolite changes characterizing GDM suggest adaptations in a number of different pathways and highlight the relevance of gut microflora disturbances in relation to the disease. The impact of diet and insulin treatments on the excreted metabolome of pregnant GDM women was measured and enabled responsive and resistant metabolic pathways to be identified, as well as side-effects of treatment i.e. metabolic changes induced by treatment and previously unrelated to the disease (including changes in the gut microflora). Furthermore, treatment duration was found to be associated to urine metabolic profile, thus emphasizing the possible future use of urine metabolomics in treatment follow-up and efficacy evaluation. Finally, a possible association of a priori urinary metabolome with future treatment requirements is reported, albeit requiring demonstration in larger cohorts. This result supports the hypothesis of different metabotypes characterizing different subjects and relating to individual response to treatment.
Conclusion
A 12-resonance metabolic signature of GDN at the time of diagnosis was identified and the evaluation of the impact of insulin and/or diet therapies enabled responsive/resistant metabolic pathways and treatment side-effects to be identified.
A common problem in metabolomics data analysis is the existence of a substantial number of missing values, which can complicate, bias, or even prevent certain downstream analyses. One of the most widely-used solutions to this problem is imputation of missing values using a k-nearest neighbors (kNN) algorithm to estimate missing metabolite abundances. kNN implicitly assumes that missing values are uniformly distributed at random in the dataset, but this is typically not true in metabolomics, where many values are missing because they are below the limit of detection of the analytical instrumentation.
Objectives
Here, we explore the impact of nonuniformly distributed missing values (missing not at random, or MNAR) on imputation performance. We present a new model for generating synthetic missing data and a new algorithm, No-Skip kNN (NS-kNN), that accounts for MNAR values to provide more accurate imputations.
Methods
We compare the imputation errors of the original kNN algorithm using two distance metrics, NS-kNN, and a recently developed algorithm KNN-TN, when applied to multiple experimental datasets with different types and levels of missing data.
Results
Our results show that NS-kNN typically outperforms kNN when at least 20–30% of missing values in a dataset are MNAR. NS-kNN also has lower imputation errors than KNN-TN on realistic datasets when at least 50% of missing values are MNAR.
Conclusion
Accounting for the nonuniform distribution of missing values in metabolomics data can significantly improve the results of imputation algorithms. The NS-kNN method imputes missing metabolomics data more accurately than existing kNN-based approaches when used on realistic datasets.
Untargeted metabolomics workflows include numerous points where variance and systematic errors can be introduced. Due to the diversity of the lipidome, manual peak picking and quantitation using molecule specific internal standards is unrealistic, and therefore quality peak picking algorithms and further feature processing and normalization algorithms are important. Subsequent normalization, data filtering, statistical analysis, and biological interpretation are simplified when quality data acquisition and feature processing are employed.
Objectives
Metrics for QC are important throughout the workflow. The robust workflow presented here provides techniques to ensure that QC checks are implemented throughout sample preparation, data acquisition, pre-processing, and analysis.
Methods
The untargeted lipidomics workflow includes sample standardization prior to acquisition, blocks of QC standards and blanks run at systematic intervals between randomized blocks of experimental data, blank feature filtering (BFF) to remove features not originating from the sample, and QC analysis of data acquisition and processing.
Results
The workflow was successfully applied to mouse liver samples, which were investigated to discern lipidomic changes throughout the development of nonalcoholic fatty liver disease (NAFLD). The workflow, including a novel filtering method, BFF, allows improved confidence in results and conclusions for lipidomic applications.
Conclusion
Using a mouse model developed for the study of the transition of NAFLD from an early stage known as simple steatosis, to the later stage, nonalcoholic steatohepatitis, in combination with our novel workflow, we have identified phosphatidylcholines, phosphatidylethanolamines, and triacylglycerols that may contribute to disease onset and/or progression.
Concerning NMR-based metabolomics, 1D spectra processing often requires an expert eye for disentangling the intertwined peaks.
Objectives
The objective of NMRProcFlow is to assist the expert in this task in the best way without requirement of programming skills.
Methods
NMRProcFlow was developed to be a graphical and interactive 1D NMR (1H & 13C) spectra processing tool.
Results
NMRProcFlow (http://nmrprocflow.org), dedicated to metabolic fingerprinting and targeted metabolomics, covers all spectra processing steps including baseline correction, chemical shift calibration and alignment.
Conclusion
Biologists and NMR spectroscopists can easily interact and develop synergies by visualizing the NMR spectra along with their corresponding experimental-factor levels, thus setting a bridge between experimental design and subsequent statistical analyses.
The chemical sensitivity of urine metabolomics analysis is greatly compromised due to the large amounts of inorganic salts in urine (NaCl, KCl), which are detrimental to analytical instrumentation, e.g. chromatographic columns or mass spectrometers. Traditional desalting approaches applied to urine pretreatment suffer from the chemical losses, which reduce the information depth of analysis.
Objectives
We aimed to test a simple approach for the simultaneous preconcentration and desalting of organic solutes in urine based on the collection of induced bursting bubble aerosols above the surface of urine samples.
Method
Bursting bubbles were generated at ambient conditions by feeding gas through an air diffuser at the bottom of diluted (200 times in ultrapure water) urine solution (50–500 mL). Collected aerosols were analyzed by the direct-infusion electrospray ionization mass spectrometry (ESI–MS).
Results
The simultaneous preconcentration (ca. 6–12 fold) and desalting (ca. six–tenfold) of organic solutes in urine was achieved by the bursting bubble sample pretreatment, which allowed ca. three-times higher number of identified urine metabolites by high-resolution MS analysis. No chemical losses due to bubbling were observed. The increased degree of MS data clustering was demonstrated on the principal component analysis of data sets from the urine of healthy people and from the urine people with renal insufficiency. At least ten times higher sensitivity of trace drug detection in urine was demonstrated for clenbuterol and salbutamol.
Conclusion
Our results indicate the high versatility of bubble bursting as a simple pretreatment approach to enhance the chemical depth and sensitivity of urine analysis. The approach could be attractive for personalized medicine as well as for the diagnostics of renal disorders of different etiology (diabetic nephropathy, chronic renal failure, transplant-associated complications, oncological disorders).
Graphical Abstract
Urine desalting and preconcentration in bursting bubbles.
Experiments in metabolomics rely on the identification and quantification of metabolites in complex biological mixtures. This remains one of the major challenges in NMR/mass spectrometry analysis of metabolic profiles. These features are mandatory to make metabolomics asserting a general approach to test a priori formulated hypotheses on the basis of exhaustive metabolome characterization rather than an exploratory tool dealing with unknown metabolic features.
Objectives
In this article we propose a method, named ASICS, based on a strong statistical theory that handles automatically the metabolites identification and quantification in proton NMR spectra.
Methods
A statistical linear model is built to explain a complex spectrum using a library containing pure metabolite spectra. This model can handle local or global chemical shift variations due to experimental conditions using a warping function. A statistical lasso-type estimator identifies and quantifies the metabolites in the complex spectrum. This estimator shows good statistical properties and handles peak overlapping issues.
Results
The performances of the method were investigated on known mixtures (such as synthetic urine) and on plasma datasets from duck and human. Results show noteworthy performances, outperforming current existing methods.
Conclusion
ASICS is a completely automated procedure to identify and quantify metabolites in 1H NMR spectra of biological mixtures. It will enable empowering NMR-based metabolomics by quickly and accurately helping experts to obtain metabolic profiles.
