A copy of this work was available on the public web and has been preserved in the Wayback Machine. The capture dates from 2019; you can also visit <a rel="external noopener" href="http://metabolomics2016.org/images/Poster-Abstracts-6-22-16.pdf">the original URL</a>. The file type is <code>application/pdf</code>.
<i title="Elsevier BV">
<a target="_blank" rel="noopener" href="https://fatcat.wiki/container/egeop3uvv5cphmwoyxjyrjrc3u" style="color: black;">Food Research International</a>
Submission: Targeted or non-targeted metabolomic analyses are, as all in vitro or in vivo measurements, prone to technical errors. These errors can occur at several states of sample preparation and may lead to, in worst cases, nonsensical data which can't be rescued even by the most sophisticated biostatistical algorithms. In less severe cases these technical errors are, if not detected early on, responsible for lengthy and more extensive data analysis. To minimize this risk we created the<span class="external-identifiers"> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1016/j.foodres.2018.06.016">doi:10.1016/j.foodres.2018.06.016</a> <a target="_blank" rel="external noopener" href="https://www.ncbi.nlm.nih.gov/pubmed/30131118">pmid:30131118</a> <a target="_blank" rel="external noopener" href="https://fatcat.wiki/release/3ha7tt3hnneofg5wz645ekssse">fatcat:3ha7tt3hnneofg5wz645ekssse</a> </span>
more »... are package "MoOD" for R Shiny. The aim of the software is to provide for non-expert users an easy to use interface to quickly check the quality of sample preparation and measurement. The user is presented with an intuitive GUI and starts with the upload of a text file. MoOD has been developed initially for data derived from MetIDQ® containing metabolomic data of the Biocrates® p150 and p180 assays. The calculations are primarily presented in two tabs of the output pane: 1) general and 2) detailed information of data composition of marker metabolites. The marker metabolites are representative of the full metabolite spectrum of the two assays and grouped by their chemical properties. The general information pane provides the user with QQ-plots of the metabolites combined with outlier detection by boxplots of log2 transformed and original scale data. From this pane it is directly visible if outliers in the data have been detected. The detailed information tab provides the user with detailed outlier detection for each marker metabolite. For now, MoOD is restricted to targeted metabolomic measurements with p150 and p180 kits. However, the code is very adaptable and other types of targeted or even non-targeted metabolomics -as well as more biostastistics for the whole dataset -are going to be included in future versions of the software. Abstract Submission: LC/MS is the most common experimental setup for untargeted metabolomics. In a typical LC/MS experiment, compounds are first separated in the chromatography and then ionized at the spectrometer, where they are analyzed. In this process, molecules from the same metabolite can undergo different transformations, usually incorporating or losing precise molecular moieties. Each of these newly formed molecules or adducts of the same metabolite produces a different signal in the spectrometer, increasing the complexity of the analysis. Correct interpretation of these signals is crucial for a rigorous and accurate metabolomics experiment. To aid in this interpretation we have developed AddClique, a network based algorithm that is able to systematically discriminate adducts from the same metabolite from those from another metabolite, and then identify them. AddClique first computes correlations between peaks of the chromatogram to obtain the probability that those peaks correspond to adducts of the same molecule. With these probabilities AddClique builds a network, with adducts as nodes and probabilities as links. Then, AddClique identifies groups of fully connected components (cliques), as these are the most probable adducts of the same metabolite. Finally AddClique identifies the adducts with their masses as well as the metabolite. We tested our algorithm with datasets of increasing complexity: from single molecule experiments to a complex biological sample. We show that AddClique is a valid tool for the identification of adducts and can be incorporated to the regular workflow in the analysis of metabolomics data. Abstract Submission: N-acylethanolamines, an endogenous lipid mediator in various animals, is amide compound group with a chemical structure condensed from long chain fatty acid and ethanol amide. In this chemical group, Narachidonoylethanolamine (anandamide) is arachidonic acid which was discovered as an endogenous ligand of cannabinoid receptor (CB1) and provides cannabinoid-like biological activities, such as analgesia and hypotensive effects. In the same class, N-palmitoylethanolamine provides antiinflammatory and analgesic action, and N-oleoylethanolamine function as anti-inflammatory. Thus, Nacylethanolamines are considered as potential biomarkers for various diseases, however, human plasma metabolite database named NIST Standard Reference Material for Human Plasma (SRM 1950) does not include this metabolite. Here, we developed lipid profiling methods using liquid chromatographytandem mass spectrometry (LC-MS/MS) to profile these metabolites in Tsuruoka metabolomics study, a large prospective cohort study in Japan, and cataloged the concentration of the N-acylethanolamine metabolites. Abstract Title: Metabolic changes in prefrontal cortex of humans, chimpanzees and macaques during postnatal development. Abstract Submission: Human evolution is characterized by changes in brain size and organization. It has long been suggested that these structural changes are linked to modifications in brain metabolism. In our study we assessed metabolic features unique to human brain by measuring intensity of 5750 metabolic peaks detected using liquid chromatography coupled with mass spectrometry in positive ([+] LC-MS) and negative ([-] LC-MS) ionization mode in a specific area of dorsolateral prefrontal cortex of 40 humans, 40 chimpanzees, and 40 rhesus monkeys. In all species samples span the entire development: from 2 days to 61 years in humans, from newborn to 42 years in chimpanzees, and from 13.6 weeks post-conception to 21 years in macaques. In order to assess metabolite concentration changes after death, we further conducted measurements in additional postmortem samples of two rhesus macaques. We show that the brain metabolome undergoes substantial changes with approximately 75% of detected metabolites showing significant concentration changes with age. Notably, 80% of these metabolic changes differ significantly among species with approximately two-fold greater number of changes taking place on the human evolutionary lineage compared to the chimpanzee lineage. The excess of human-specific divergence was not distributed uniformly across lifespan, but peaked between 40 and 55 years of human age. Coupling of human-specific metabolic changes with corresponding changes in mRNA abundance signed out TCA cycle and arginine and proline metabolism, as well as number of other pathways, as the ones enriched in human-specific changes. Abstract Title: Assessing the penetration of antimicrobials into Pseusomonas aeruginosa biofilms using ToF-SIMS Abstract Submission: Cells that exist as part of a biofilm are much less susceptible to antimicrobials than are planktonic cells. Biofilms commonly grow on medical equipment and inserts such as catheters, and are often extremely difficult to remove. Pseudomonas aeruginosa biofilms grow in the lungs of cystic fibrosis patients, causing chronic infections. We have used Time-of-Flight Secondary Ion Mass Spectrometry (ToF-SIMS) to compare the penetration of three clinically relevant antimicrobials (tobramycin, ciprofloxacin and gallium) into P. aeruginosa biofilms. ToF-SIMS combines high spatial resolution (up to 300 nm) with high sensitivity. We grew P. aeruginosa biofilms on indium tin oxide (ITO)-coated glass, as a flat, electrically conductive substrate is preferred for ToF-SIMS analysis. We confirmed biofilm formation on this substrate with SEM. ITO coated glass pieces were incubated in liquid P. aeruginosa culture (PA14 retS) for 96 hours to allow biofilm formation, then removed before the addition of a mixture of antimicrobials with a range of exposure times. Samples were lyophilized before MSI analysis. MS imaging was carried out at a submicron resolution (with a 0-800 m/z range) using a TOF.SIMS 5 secondary ion mass spectrometer (IONTOF). This instrument can operate in two modes. The first is a static mode for surface MS imaging (X-Y axis) using a Bi3+ ion beam, causing minimal damage to the sample surface. The second is a dynamic, destructive mode for depth profiling, using an argon cluster primary ion beam to sputter material from the sample, enabling the acquisition of sequential MS data on the Z-axis. By analysing pure antibiotic standards, and depth profiling through the biofilm to the substrate, we were able to obtain and compare profiles of antimicrobial penetration into the biofilm, as well as identifying endogenous chemical gradients within the biofilm itself. Abstract Submission: Chemometric bioinformatics is a concept where chemometric methods, i.e. design of experiments (DOE) and multivariate analysis by means of projections (MVA), are applied in solving, and developing new tools for, bioinformatic problems. Our ambition is to increase the information output from metabolomics data by further developments and refinements of chemometric methods and strategies. In this presentation the use of DOE and MVA methods in metabolomics studies will be critically discussed and suggestions of novel approaches utilizing the uniqueness of the chemometric methodology will be given. Special emphasis will be on how to increase the sensitivity of biomarker or biomarker pattern detection as well as how to obtain a more comprehensive and correct interpretation of the systematic metabolic pattern changes and how to deal with this statistically. Examples from studies including matched or dependent samples will be given and discussed. In addition, the impact and use of orthogonal systematic variation in metabolomics data will be given specific attention for defining the significance of metabolites or patterns thereof. All presented strategies will be exemplified in real clinical studies ranging from early disease detection to treatment response and controlled interventions. Abstract Title: Integrated gene expression and metabolomics analysis defines molecular cancer signatures Abstract Submission: Cancer remains a leading cause of death. Every year, over 14 million people are diagnosed with cancer worldwide and over 8 million people will die of the disease. With the strong potential of using metabolites as guides for predicting early diagnosis, prognosis, and treatment outcome, metabolomics has gained momentum in the recent years. Oftentimes though, the interpretation of metabolomics profiles, especially in large untargeted studies, is challenging and the biological mechanisms underlying cancer-specific profiles are frequently unknown. Yet understanding the regulation of metabolic enzymes involved in producing cancer phenotypes is critical and could facilitate the search for novel therapeutic targets. To this end, a global approach that integrates gene expression with metabolite measurements is proposed. Importantly, this approach could be extended to the integration of other omics data (e.g. proteomics, epigenomics). Highly correlated gene:metabolite pairs may reveal genes that directly or indirectly affect the abundances of metabolites related to a specific cancer phenotype. We hypothesize that gene:metabolite correlations are associated with different phenotypes and that these pairs may reveal key biological mechanisms that affect different phenotypes. Leveraging the public NCI-60 cell line data, we applied a linear model, m = g + c + g:c, where m are metabolite abundances, g are gene expression values, c is cancer type (e.g. leukemia, prostate), and g:c is the interaction between gene expression and cancer type. A statistically significant interaction p-value indicates that the corresponding gene:metabolite pair is highly correlated in one cancer type but not the other, and that the gene:metabolite pair is cancer-type specific. Applying these models to all possible gene:metabolite pairs (N=4,640,646) to compare different cancer subtypes in the NCI-60 cell lines yielded potentially relevant cancer-type specific gene:metabolite pairs. Results of this approach and comparison with other transcriptomics/metabolomics approaches will be discussed here. Abstract Submission: Parametric Time Warping distorts the time axis of a chromatogram in order to maximize the overlap with a reference. The original version worked on complete time profiles only. However, often only peak positions are available, e.g., when comparing to data bases of known compounds. Since 2015, the R package ptw implementing this technique now also supports aligning peak lists , e.g., coming from XCMS peak picking. It is shown that the technique is up to several orders of magnitude faster, since many irrelevant data points are no longer taken into account. An added benefit is that the careful preprocessing (in particular baseline subtraction) is nowhere near as important when aligning peaks -the peak picking method has made sure only the relevant information is available for the alignment algorithm. A final advantage of the new method is that wider time windows no longer take more time, and that this practically means that one does not have to fine-tune the algorithm for a particular data set. We show examples of the use of ptw for a number of metabolomics experiments, using both LC-MS and LC-DAD data. This methodology will become ever more important with the increased access to FIAR data, allowing the simple and fast set up of restricted yet powerful time warping models. The R package, including demonstration data, is available from CRAN.  R. Wehrens, T. Bloemberg and P. Eilers: Fast Parametric Time Warping of Peak Lists, Bioinformatics (2015) Abstract Submission: The need for reproducible and comparable results is of increasing importance in non-targeted metabolomics, especially when differences between experimental groups are small. Liquid chromatography -mass spectrometry (LC-MS) spectra are often acquired batch-wise so that necessary calibrations and cleaning of the instrument can take place. However, this may introduce further sources of variation, such as differences in the conditions under which the acquisition of individual batches is performed. Quality control (QC) samples are frequently employed as a means of both judging and correcting this variation. However, the use of QC samples can lead to problems. The non-linearity of the response can result in substantial differences between the recorded intensities of the QCs and experimental samples, making the required adjustment difficult to predict. Furthermore, changes in the response profile between one QC and the next cannot be accounted for and QC based correction can actually exacerbate the problems by introducing artificial differences. We introduce a "background correction" method that utilises all experimental samples to estimate the variation over time rather than relying on the QC samples alone. The method is compared with standard QC correction in terms of the reduction in differences between replicate samples and the potential to highlight differences between experimental groups previously hidden by instrumental variation. Abstract Submission: Modelling of metabolomics data in targeted and untargeted approach is usually aimed at searching for potential disease indicators. Widely adopted multivariate approaches (PCA, PLS-DA, OPLS) however, are design to reduce dimensionality and perform multivariate analysis providing limited information on the usefulness of metabolites as potential disease indicators. Alternatively, metabolomics data can be fitted to the pharmacokinetics-based models allowing for more mechanistic characterization of collected data via natural incorporation of relationships between patient characteristics (age, gender) and model parameters to make inference about the disease status. In this work we proposed the concept of pharmacokinetics-driven fully Bayesian approach to (i) model nucleosides/creatinine concentration ratios in a function of age, gender and case/control status and to further (ii) assess the posterior prediction of probability of cancer occurrence in an individual subject. The data set was randomly divided into training and validation set. Non-informative priors were used for modelling purposes. The model performance was evaluated using a posteriori predictive check. The validation set was used to evaluate the probability of cancer development for individuals with known nucleosides/creatinine concentration ratios, age and gender based on the developed model. The accuracy of classification was summarized by area under the ROC(AUROC), sensitivity and specificity. As a consequence of cancer, methylthioadenosine concentration increased by a factor of exp(1.82)((exp(1.33)-exp(2.47)). Age influences nucleosides/creatinine concentration ratios for all nucleosides in the same direction which is likely caused by decrease of creatinine clearance with age. The individual a posteriori prediction of disease expressed via AUROC was 0.6 (0.51-0.69) with sensitivity and specificity of 0.63 (0.46-0.76) and 0.58 (0.45-0.68), respectively suggesting limited usefulness of nucleosides in predicting patient's disease status in this population. The widely adopted pharmacokinetics-based approach in drug development can be used to describe metabolomics data. Abstract Title: Statistical considerations in handling unwanted variation in large-scale metabolomics studies Abstract Submission: Large-scale metabolomics studies often require the use of multiple analytical platforms, batches of samples, and laboratories, any of which can introduce a component of unwanted variation. In addition, every experiment is subject to within-platform and other experimental variation, which often includes unwanted biological variation. A notable difficulty in capturing the component of unwanted variation is that both the biological and experimental unwanted variation can be unobserved as well as observed. Although the removal of this unwanted variation is a vital step in the analysis of metabolomics data, it is considered a gray area in which there is a recognized need to develop a better understanding of the procedures and statistical methods required to achieve statistically relevant optimal biological outcomes.  De Livera, A. M et al (2012). Normalizing and integrating metabolomics data. Abstract Submission: Background: Huntington's disease (HD) is a fatal autosomal-dominant neurodegenerative disorder affecting approximately 3-10 people per 100,000 in the Western World. The median-age of onset is 40 years, with death typically following 15-20 years later. Objectives: In this study we aimed to biochemically profile post-mortem (PM) human brain excised from the frontal lobe and striatum of HD sufferers (n=14) and compared their profiles with controls (n=14). Our overarching goals were to identify potential central biomarkers of HD whilst providing an insight into the molecular basis of the disease by better understanding the biochemistry behind its progression. Methods: We employed LC-LTQ-Orbitrap-MS for the global metabolite profiling of the polar metabolome from the striatum and frontal lobe regions. Results: A total of 5,579 and 5,880 features were detected in the frontal lobe and striatum, respectively. An ROC curve combining two spectral features from frontal lobe had an AUC value of 0.916 (0.794 -1.000) and following statistical cross-validation had an 83% predictive accuracy for HD. Similarly, two striatum biomarkers gave an ROC AUC of 0.935 (0.806 -1.000) and following statistical cross-validation predicted HD with 91.8% accuracy. A range of metabolite disturbances were evident including but-2enoic acid and uric acid which were altered in both frontal lobe and striatum. A total of 7 biochemical pathways (3 in the frontal lobe and 4 in the striatum) were significantly altered as a result of HD pathology. Conclusion: This study highlights the utility of high-resolution metabolomics for the study of HD. Further characterization of the brain metabolome could lead to the identification of new biomarkers and novel treatment strategies for HD. Abstract Title: A metabolomics approach for identification of novel biomarkers of chicken consumption Abstract Submission: A metabolomics approach for identification of novel biomarkers of chicken consumption Background: Numerous studies have highlighted associations between meat intake and health outcomes. However, reliable and accurate dietary assessment methods are essential to confirm these proposed associations. Objective: To identify and validate novel dietary biomarkers of chicken intake using a metabolomics approach. Materials and methods: Urine samples from NutriTech Food Intake Study were used where volunteers consumed increasing amounts of chicken for three consecutive weeks. For example, females consumed 86 g/d in week 1, 176 g/d in week 2, and 308 g/d in week 3. The samples were analysed by a 600-MHz Varian NMR spectrometer and multivariate data analysis was performed with Simca-P software. Estimation of mean, standard deviation, and significant difference were calculated by SPSS. The putative biomarker was validated in a free-living population from National Adult Nutrition Survey (NANS). Receiver operating characteristic (ROC) was performed to evaluate the sensitivity and specificity of the biomarker. Result: The application of PCA and PLS-DA models of postprandial and fasting urine samples revealed good separations between high and low chicken intake. Discriminatory regions were identified from the S-line plot. Examination of the NMR profiles led to the identification of guanidoacetate as a metabolite increased following chicken consumption. The concentrations of guanidoacetate in fasting urine samples significantly increased with increasing chicken intake (P < 0.001), and were higher compared to the red meat group. ROC analysis to assess the classification ability of guanidoacetate between red meat and chicken groups represented a specificity and sensitivity of 0.90 and 0.98, respectively. The biomarker was confirmed in NANS cohort where chicken consumers had significantly higher concentrations of guanidoacetate (P < 0.001), compared with non-consumers. Conclusion: Guanidoacetate was successfully identified and confirmed as a biomarker of chicken intake by a metabolomics-based approach. Abstract Submission: Untargeted metabolomics is a powerful phenotyping tool for better understanding biological mechanisms involved in human pathology development and identifying early predictive biomarkers. However, this approach generates massive and complex data that need appropriate analyses to extract biologically meaningful information (Xi, 2014). In this context, this work consists in designing a workflow using knowledge discovery and data mining methodologies to propose advanced solutions for predictive biomarker discovery. Data were collected from a mass spectrometry-based untargeted metabolomic approach performed on subjects from a case/control study within the GAZEL French population-based cohort. Different feature selection approaches were applied either on the original metabolomic dataset or on reduced subsets. The strategy was focused on evaluating a combination of numeric-symbolic approaches for feature selection with the objective of obtaining the best combination of metabolites, producing an effective and accurate predictive model. Relying first on numerical approaches, and especially on machine learning methods (SVM and RF) and on univariate statistical analyses (ANOVA), a comparative study was performed on the original metabolomic dataset and reduced subsets. The best k-features obtained with different scores of importance from the combination of these different approaches were compared and allowed determining the variable stabilities using Formal Concept Analysis. The results revealed the interest of RF-Gini combined with ANOVA for feature selection as these two complementary methods allowed selecting the 48 best candidates for prediction. Using linear logistic regression strategy on this reduced dataset enabled us to obtain the best performances in prediction with a model including 5 top variables. Therefore, these results highlighted the interest of feature selection methods and the importance of working on reduced datasets for the identification of predictive biomarkers issued from untargeted metabolomics data. These data mining methods are essential tools to deal with massive datasets and contribute to elucidate complex phenomena associated with chronic disease development. Abstract Submission: Tandem Mass spectrometry (MS/MS) is a widely used approach to annotate and identify metabolites in complex biological samples. The importance of assessing the contribution of the precursor peak within an isolation window for MS/MS has been previously detailed in proteomics but to date there has been little attention paid to this data-processing technique in metabolomics. Here we present msPurity, a vendor independent R package for liquid chromatography (LC) and direct infusion (DI) MS/MS that calculates a simple metric to describe the contribution of the selected precursor peak to the population of isolated ions. What we call here "precursor purity" is calculated as per the Michalski approach (intensity of selected precursor divided by total intensity of the isolation window) with the exception that the metric is interpolated at the time of the MS/MS scan. The package was applied to Data Dependent Acquisition (DDA) based MS/MS metabolomics datasets derived from three metabolomic data repositories. For the ten LC-MS/MS DDA datasets with isolation windows less-than or equal to 1 +/-Da the median precursor purity score ranged from 0.61 to 0.85. The R package was also used to predict the precursor purity from an LC-MS dataset of a complex biological sample (Daphnia magna) using 0.5 +/-Da isolation windows, where the median predicted precursor purity score of the full width half maximum (FWHM) for all XCMS-determined features was 0.51. We therefore demonstrate that for complex samples there will be a large number of metabolites where traditional DDA approaches may struggle to provide reliable annotations. Abstract Submission: Introduction: Since 2013, MetaboHUB, the French infrastructure for metabolomics and fluxomics, provides tools and services to academic research teams and industrial partners in the fields of health, nutrition, agriculture, environment and biotechnology. One of its last developments is PeakForest.org, the MetaboHUB's reference spectral database. The first version of this tool already provides basic functionalities like a spectral database of standard compounds and querying / visualization tools plugged on it. The final version will include metabolome annotations of biological reference matrices. PeakForest Database: the current release provides templates and a user-friendly interface in order to collect high quality spectral data (peak lists and metadata) from proton 1D-NMR and LC-MS instruments.
<a target="_blank" rel="noopener" href="https://web.archive.org/web/20190429100208/http://metabolomics2016.org/images/Poster-Abstracts-6-22-16.pdf" title="fulltext PDF download" data-goatcounter-click="serp-fulltext" data-goatcounter-title="serp-fulltext"> <button class="ui simple right pointing dropdown compact black labeled icon button serp-button"> <i class="icon ia-icon"></i> Web Archive [PDF] </button> </a> <a target="_blank" rel="external noopener noreferrer" href="https://doi.org/10.1016/j.foodres.2018.06.016"> <button class="ui left aligned compact blue labeled icon button serp-button"> <i class="external alternate icon"></i> elsevier.com </button> </a>