Untargeted Metabolomics Data Research Articles

Altered urinary tryptophan metabolites in alcohol-associated liver disease.

Alcohol-associated liver disease (ALD) leads to millions of deaths worldwide annually. A few potential biomarkers of ALD have been discovered through metabolomic or proteomic analysis. Tryptophan (Trp), one of nine essential amino acids, has been extensively studied and shown to play significant roles in many mammalian physiological processes. However, Trp metabolism changes in ALD are not yet fully understood. Whereas urine is an abundant and non-invasive source for disease biomarker discovery the current study investigated whether the abundance of Trp metabolites in the urine of ALD patients differs from that of healthy subjects. We also examined whether, if present in ALD, changes in urinary Trp metabolites can serve as markers for differentiating between mild/moderate and severe ALD. We quantified the concentration of Trp and its metabolites in urine samples of healthy controls (n = 18), patients with mild or moderate alcohol-related liver injury (non-severe ALD; n = 21), and patients with severe alcohol-associated hepatitis (severe AH; n = 25) using both untargeted and targeted metabolomics. Eighteen Trp metabolites were identified and quantified from the untargeted metabolomics data. We developed a targeted metabolomics method to quantify the Trp and its metabolites and quantified 17 metabolites from the human urine samples. The data acquired in the untargeted and targeted platforms agreed and showed that the Trp concentration is not affected by the severity of ALD. However, the abundance of 10 Trp metabolites was correlated with the model for end-stage liver disease (MELD) score, with the abundance of nine metabolites significantly different between the healthy control and ALD patient groups. We found that Trp metabolism differs between ALD patients and healthy controls even though the concentration of Trp was not affected. Two Trp metabolites, quinolinic acid and indoxyl sulfate, correlate highly with the severity of ALD.

Precise prediction of metabolites patterns using machine learning approaches in distinguishing honey and sugar diets fed to mice

As a natural sweetener produced by honey bees, honey was recognized as being healthier for consumption than table sugar. Our previous study also indicated thatmetaboliteprofiles in mice fed honey and mixedsugardiets aredifferent. However, it is still noteworthy about the batch-to-batch consistency of the metabolic differences between two diet types. Here, the machine learning (ML) algorithms were applied to complement and calibrate HPLC-QTOF/MS-based untargeted metabolomics data. Data were generated from three batches of mice that had the same treatment, which can further mine the metabolite biomarkers. Random Forest and Extra-Trees models could better discriminate between honey and mixed sugar dietary patterns under five-fold cross-validation. Finally, SHapley Additive exPlanations tool identified phosphatidylethanolamine and phosphatidylcholine as reliable metabolic biomarkers to discriminate the honey diet from the mixed sugar diet. This study provides us new ideas for metabolomic analysis of larger data sets.

IpaPy2: Integrated Probabilistic Annotation (IPA) 2.0 - an improved Bayesian-based method for the annotation of LC-MS/MS untargeted metabolomics data.

The Integrated Probabilistic Annotation (IPA) is an automated annotation method for LC-MS-based untargeted metabolomics experiments, that provides statistically rigorous estimates of the probabilities associated with each annotation. Here we introduce ipaPy2, a substantially improved and completely refactored Python implementation of the IPA method. The revised method is now able to integrate tandem MS fragmentation data, which increases the accuracy of the identifications. Moreover, ipaPy2 provides a much more user-friendly interface, and isotope peaks are no longer treated as individual features, but integrated into isotope fingerprints, greatly speeding up the calculations. The method has also been fully integrated with the mzMatch pipeline, so that the results of the annotation can be explored through the newly developed PeakMLViewerPy tool available at https://github.com/UoMMIB/PeakMLViewerPy. The source code, extensive documentation and tutorials are freely available on GitHub at https://github.com/francescodc87/ipaPy2. Supplementary data are available at Bioinformatics online.

The adipose-derived stem cell peptide ADSCP2 alleviates hypertrophic scar fibrosis via binding with pyruvate carboxylase and remodeling the metabolic landscape.

The purpose of this study was to investigate the function and mechanism of a novel peptide derived from adipose-derived stem cell-conditioned medium (ADSC-CM). Mass spectrometry was applied to identify expressed peptides in ADSC-CM obtained at different time points. The cell counting kit-8 assay and quantitative reverse transcription polymerase chain reactions were performed to screen the functional peptides contained within ADSC-CM. RNA-seq, western blot, a back skin excisional model of BALB/c mice, the peptide pull-down assay, rescue experiments, untargeted metabolomics, and mixOmics analysis were performed to thoroughly understand the functional mechanism of selected peptide. A total of 93, 827, 1108, and 631 peptides were identified in ADSC-CM at 0, 24, 48, and 72 h of conditioning, respectively. A peptide named ADSCP2 (DENREKVNDQAKL) derived from ADSC-CM inhibited collagen and ACTA2 mRNAs in hypertrophic scar fibroblasts. Moreover, ADSCP2 facilitated wound healing and attenuated collagen deposition in a mouse model. ADSCP2 bound with the pyruvate carboxylase (PC) protein and inhibited PC protein expression. Overexpressing PC rescued the reduction in collagen and ACTA2 mRNAs caused by ADSCP2. Untargeted metabolomics identified 258 and 447 differential metabolites in the negative and positive mode, respectively, in the ADSCP2-treated group. The mixOmics analysis, which integrated RNA-seq and untargeted metabolomics data, provided a more holistic view of the functions of ADSCP2. Overall, a novel peptide derived from ADSC-CM, named ADSCP2, attenuated hypertrophic scar fibrosis in vitro and in vivo, and the novel peptide ADSCP2 might be a promising drug candidate for clinical scar therapy.

An integrated deep learning framework for the interpretation of untargeted metabolomics data.

Untargeted metabolomics is gaining widespread applications. The key aspects of the data analysis include modeling complex activities of the metabolic network, selecting metabolites associated with clinical outcome and finding critical metabolic pathways to reveal biological mechanisms. One of the key roadblocks in data analysis is not well-addressed, which is the problem of matching uncertainty between data features and known metabolites. Given the limitations of the experimental technology, the identities of data features cannot be directly revealed in the data. The predominant approach for mapping features to metabolites is to match the mass-to-charge ratio (m/z) of data features to those derived from theoretical values of known metabolites. The relationship between features and metabolites is not one-to-one since some metabolites share molecular composition, and various adduct ions can be derived from the same metabolite. This matching uncertainty causes unreliable metabolite selection and functional analysis results. Here we introduce an integrated deep learning framework for metabolomics data that take matching uncertainty into consideration. The model is devised with a gradual sparsification neural network based on the known metabolic network and the annotation relationship between features and metabolites. This architecture characterizes metabolomics data and reflects the modular structure of biological system. Three goals can be achieved simultaneously without requiring much complex inference and additional assumptions: (1) evaluate metabolite importance, (2) infer feature-metabolite matching likelihood and (3) select disease sub-networks. When applied to a COVID metabolomics dataset and an aging mouse brain dataset, our method found metabolic sub-networks that were easily interpretable.

365-OR: A Machine-Learning Approach to Estimate Insulin Resistance from Untargeted Metabolomics Data

Insulin resistance (IR) is constantly invoked as a contributor to numerous common diseases, e.g. diabetes, cardiovascular disease and Alzheimer’s disease; however, with the exception of a narrow segment of diabetes-related studies, IR is infrequently measured and rarely measured well. Our goal was to develop IR estimation models for the dynamic measure of insulin sensitivity (SI) and the basal measure of homeostatic model assessment of IR (HOMAIR) from metabolomics data combined with minimal clinical data, e.g. age, sex, BMI. Least Absolute Shrinkage and Selection Operator (LASSO) and Elastic Net regression were used to build estimation models using 1274 metabolites, prioritized by trait associations and collected in the Insulin Resistance Atherosclerosis Family Study (IRASFS). Three metabolite transformations were implemented (inverse normal transformation, standardization, and Box Cox transformation) to account for distributional differences and tested to optimize estimation. The analysis was performed in one Mexican American (MA) recruitment site (San Luis Valley, CO (SLV), N=450) and tested in another MA recruitment site (San Antonio, TX (SA), N=473). In addition, the two MA recruitment sites were combined and estimation models tested in the African American recruitment site (AA; Los Angeles, California, N=495) to assess generalizability. Estimated SI was correlated with empiric SI in the SA (r2=0.77) and AA (r2=0.74) testing datasets. Further, we tested estimated SI for association with BMI, low-density lipoprotein cholesterol and triglycerides with results consistent with those observed for empiric SI. The same framework was used to estimate HOMAIR, and yielded similar results. In summary, we have developed methods for estimating multiple measures of IR from a single blood sample. This approach will greatly expand our ability to provide meaningful insight into the role of IR in disease, supporting the potential for application to a wide-range of biomedical studies. Disclosure N.Allred: None. D.W.Bowden: None. F.Hsu: None. S.Chen: None. M.Ng: None. M.O.Goodarzi: Advisory Panel; Nestlé Health Science, Other Relationship; Nestlé Health Science. J.I.Rotter: None. L.E.Wagenknecht: None. M.Bancks: None. R.N.Bergman: Consultant; Lilly, ReCor Medical, Inc., Research Support; AstraZeneca. Funding National Institutes of Health (DK118062)

137-OR: Metabolite Ratios and Development of Type 1 Diabetes

Ratios between metabolites more directly reflect flux through chemical reactions than individual metabolites. Variation in metabolites and their ratios are determined, in part, by genetic factors. Thus, genetic loci associated with variation in metabolite ratios may help disentangle genetic from non-genetic processes captured by the metabolome. We integrated metabolite ratios with genetics to predict changes preceding type 1 diabetes (T1D) in The Environmental Determinants of Diabetes in the Young (TEDDY) study. Quantitative trait loci (QTLs) were identified for 132 metabolite ratios previously associated with progression from islet autoimmunity (IA) to T1D. Ratios were calculated from untargeted metabolomics data available at the time of seroconversion to IA for 338 cases and 921 matched controls in TEDDY. Genotyping was performed using a genome-wide array with custom content with imputation to the TOPMed reference panel. Ratio-SNP associations were estimated using Matrix eQTL software, adjusting for case-control status and genetic relatedness. Genome-wide significant association (P < 4.9×10-9) was observed at 202 independent loci, involving 49 metabolite ratios. Twenty-six loci (12.9%) involved ratios between polyunsaturated fatty-acids (PUFAs) and fatty acid esters of hydroxy fatty acids (FAHFAs), a novel class of bioactive lipids with antidiabetic and anti-inflammatory effects. Variation in the metabolite ratio of eicosadienoic acid to FAHFA (18:1/O-15:1) was significantly determined by a site on chromosome 3 (ITPR1); ITPR1 encodes a receptor integral to calcium signaling and glucose-induced insulin secretion. Other genetic loci contributing to variation of PUFA:FAHFA ratios reside in genes controlling inflammatory and innate immune response (e.g., TSPO, TTLL12, NCAM1) and apoptosis (PPM1L). These novel metabolite ratio QTLs, some with known association with T1D risk, expand potential biomarkers of T1D risk, with PUFA:FAHFA ratios representing candidate FAHFA biosynthesis genes. Disclosure R.K. Johnson: None. A. Romero: None. E. Willems: Employee; Amgen Inc. S.S. Rich: None. M. Rewers: Research Support; Provention Bio, Juvenile Diabetes Research Foundation (JDRF), Hemsley Charitable Trust, Dexcom, Inc., Janssen Research & Development, LLC. O. Fiehn: None. J.M. Norris: None. Funding National Institute of Diabetes and Digestive and Kidney Diseases (R03DK127472); National Institute of Allergy and Infectious Diseases; National Institute of Child Health and Human Development; National Institute of Environmental Health Sciences; JDRF; Centers for Disease Control and Prevention

PeakDetective: A Semisupervised Deep Learning-Based Approach for Peak Curation in Untargeted Metabolomics.

Peak-detection algorithms currently used to process untargeted metabolomics data were designed to maximize sensitivity at the sacrifice of selectively. Peak lists returned by conventional software tools therefore contain a high density of artifacts that do not represent real chemical analytes, which, in turn, hinder downstream analyses. Although some innovative approaches to remove artifacts have recently been introduced, they involve extensive user intervention due to the diversity of peak shapes present within and across metabolomics data sets. To address this bottleneck in metabolomics data processing, we developed a semisupervised deep learning-based approach, PeakDetective, for classification of detected peaks as artifacts or true peaks. Our approach utilizes two techniques for artifact removal. First, an unsupervised autoencoder is used to extract a low-dimensional, latent representation of each peak. Second, a classifier is trained with active learning to discriminate between artifacts and true peaks. Through active learning, the classifier is trained with less than 100 user-labeled peaks in a matter of minutes. Given the speed of its training, PeakDetective can be rapidly tailored to specific LC/MS methods and sample types to maximize performance on each type of data set. In addition to curation, the trained models can also be utilized for peak detection to immediately detect peaks with both high sensitivity and selectivity. We validated PeakDetective on five diverse LC/MS data sets, where PeakDetective showed greater accuracy compared to current approaches. When applied to a SARS-CoV-2 data set, PeakDetective enabled more statistically significant metabolites to be detected. PeakDetective is open source and available as a Python package at https://github.com/pattilab/PeakDetective.

Integrative processing of untargeted metabolomic and lipidomic data using MultiABLER

Mass spectrometry (MS)-based untargeted metabolomic and lipidomic approaches are being used increasingly in biomedical research. The adoption and integration of these data are critical to the overall multi-omic toolkit. Recently, a sample extraction method called Multi-ABLE has been developed, which enables concurrent generation of proteomic and untargeted metabolomic and lipidomic data from a small amount of tissue. The proteomics field has a well-established set of software for processing of acquired data; however, there is a lack ofa unified, off-the-shelf, ready-to-use bioinformatics pipeline that can take advantage of and prepare concurrently generated metabolomic and lipidomic data forjoint downstream analyses. Here we present an R pipeline called MultiABLER as a unified and simple upstream processing and analysis pipeline for both metabolomics and lipidomics datasets acquired using liquid chromatography-tandem massspectrometry. The code is available via an open-source license at https://github.com/holab-hku/MultiABLER.

UHPLC-QTOF-MS-based untargeted metabolomics revealing the differential chemical constituents and its application on the geographical origins traceability of lily bulbs

Lily bulbs have been historically used as an edible and medicinal homologous plant. Identifying the geographical origins of lily bulbs produced in specific origin is of great importance since the geographical origins of lily bulbs influence their quality and price greatly. In this work, an untargeted metabolomic method based on UHPLC-QTOF-MS was established for revealing the differential chemical constituents of lily bulbs among different origins and predicting the geographical origins of them by chemometric modeling. A total of 15 differential compounds were screened and identified from untargeted metabolomic data of 50 lily bulb samples by our previously developed AntDAS software. Principal component analysis (PCA) and hierarchical cluster analysis (HCA) showed that samples from five different origins were obviously distinguished based on the differential compounds. What's more, 7 and 6 key characteristic markers were discovered by partial least squares-discriminant analysis (PLS-DA) and orthogonal partial least squares-discriminant analysis (OPLS-DA), respectively. Finally, heatmap, PLS-DA and OPLS-DA models were reconstructed based on the discovered key characteristic markers, and external validation lily bulb samples were successfully discriminated, with recognition rate of 100%. This study demonstrated that the proposed strategy has great potentials for the differentiation and identification of the geographical origins of lily bulbs.

Extending inherited metabolic disorder diagnostics with biomarker interaction visualizations

BackgroundInherited Metabolic Disorders (IMDs) are rare diseases where one impaired protein leads to a cascade of changes in the adjacent chemical conversions. IMDs often present with non-specific symptoms, a lack of a clear genotype–phenotype correlation, and de novo mutations, complicating diagnosis. Furthermore, products of one metabolic conversion can be the substrate of another pathway obscuring biomarker identification and causing overlapping biomarkers for different disorders. Visualization of the connections between metabolic biomarkers and the enzymes involved might aid in the diagnostic process. The goal of this study was to provide a proof-of-concept framework for integrating knowledge of metabolic interactions with real-life patient data before scaling up this approach. This framework was tested on two groups of well-studied and related metabolic pathways (the urea cycle and pyrimidine de-novo synthesis). The lessons learned from our approach will help to scale up the framework and support the diagnosis of other less-understood IMDs.MethodsOur framework integrates literature and expert knowledge into machine-readable pathway models, including relevant urine biomarkers and their interactions. The clinical data of 16 previously diagnosed patients with various pyrimidine and urea cycle disorders were visualized on the top 3 relevant pathways. Two expert laboratory scientists evaluated the resulting visualizations to derive a diagnosis.ResultsThe proof-of-concept platform resulted in varying numbers of relevant biomarkers (five to 48), pathways, and pathway interactions for each patient. The two experts reached the same conclusions for all samples with our proposed framework as with the current metabolic diagnostic pipeline. For nine patient samples, the diagnosis was made without knowledge about clinical symptoms or sex. For the remaining seven cases, four interpretations pointed in the direction of a subset of disorders, while three cases were found to be undiagnosable with the available data. Diagnosing these patients would require additional testing besides biochemical analysis.ConclusionThe presented framework shows how metabolic interaction knowledge can be integrated with clinical data in one visualization, which can be relevant for future analysis of difficult patient cases and untargeted metabolomics data. Several challenges were identified during the development of this framework, which should be resolved before this approach can be scaled up and implemented to support the diagnosis of other (less understood) IMDs. The framework could be extended with other OMICS data (e.g. genomics, transcriptomics), and phenotypic data, as well as linked to other knowledge captured as Linked Open Data.

Encoding LC-MS-Based Untargeted Metabolomics Data into Images toward AI-Based Clinical Diagnosis.

Liquid chromatography-mass spectrometry (LC-MS)-based untargeted metabolomics provides comprehensive and quantitative profiling of metabolites in clinical investigations. The use of whole metabolome profiles is a promising strategy for disease diagnosis but technically challenging. Here, we developed an approach, namely MetImage, to encode LC-MS-based untargeted metabolomics data into multi-channel digital images. Then, the images that represent the comprehensive metabolome profiles can be employed for developing deep learning-based AI models toward clinical diagnosis. In this work, we demonstrated the application of MetImage for clinical screening of esophageal squamous cell carcinoma (ESCC) in a clinical cohort with 1104 participants. A convolutional neuronal network-based AI model was trained to distinguish ESCC screening positive and negative subjects using their serum metabolomics data. Superior performances such as sensitivity (85%), specificity (92%), and area under curve (0.95) were validated in an independent testing cohort (N = 442). Importantly, we demonstrated that our AI-based ESCC screening model is not a "black box". The encoded images reserved the characteristics of mass spectra from the raw LC-MS data; therefore, metabolite identifications in key image features were readily achieved. Altogether, MetImage is a unique approach that encodes raw LC-MS-based untargeted metabolomics data into images and facilitates the utilization of whole metabolome profiles for AI-based clinical applications with improved interpretability.

Untargeted metabolomics and lipidomics to assess plasma metabolite changes in dairy goats with subclinical hyperketonemia

Subclinical hyperketonemia (SCHK) is the major metabolic disease observed during the transition period in dairy goats, and is characterized by high plasma levels of nonesterified fatty acids (NEFA) and β-hydroxybutyrate (BHB). However, no prior study has comprehensively assessed metabolomic profiles of dairy goats with SCHK. Plasma samples were collected within 1 h after kidding from SCHK goats (BHB concentration >0.8 mM, n = 7) and clinically healthy goats (BHB concentration <0.8 mM, n = 7) with similar body condition score (2.75 ± 0.15, mean ± standard error of the mean) and parity (primiparous). A combination of targeted and untargeted mass spectrometric approaches was employed for analyzing the various changes in the plasma lipidome and metabolome. Statistical analyses were performed using the GraphPad Prism 8.0, SIMCA-P software (version 14.1), and R packages (version 4.1.3). Plasma aminotransferase, nonesterified fatty acids, and BHB concentrations were greater in the SCHK group, but plasma glucose concentrations were lower. A total of 156 metabolites and 466 lipids were identified. The analysis of untargeted metabolomics data by principal component analysis and orthogonal partial least squares discriminant analysis revealed a separation between SCHK and clinically healthy goats. According to the screening criteria (unpaired t-test, P < 0.05), 30 differentially altered metabolites and 115 differentially altered lipids were detected. Pathway enrichment analysis identified citrate cycle, alanine, aspartate and glutamate metabolism, glyoxylate and dicarboxylate metabolism, and phenylalanine metabolism as significantly altered pathways. A greater concentration of plasma isocitric acid and cis-aconitic acid levels was observed in SCHK goats. In addition, AA such as lysine and isoleucine were greater, whereas alanine and phenylacetylglycine were lower in SCHK dairy goats. Dairy goats with SCHK also exhibited greater oleic acid, acylcarnitine, and phosphatidylcholine and lower choline and sphingomyelins. Acylcarnitines, oleic acid, and tridecanoic acid displayed positive correlations with several lipid species. Alanine, hippuric acid, and histidinyl-phenylalanine were negatively correlated with several lipids. Overall, altered metabolites in SCHK dairy goats indicated a more severe degree of negative energy balance. Data also indicated an imbalance in the tricarboxylic acid (TCA) cycle, lipid metabolism, and AA metabolism. The findings provide a more comprehensive understanding of the pathogenesis of SCHK in dairy goats.

Mechanistic Understanding of the Discrepancies between Common Peak Picking Algorithms in Liquid Chromatography-Mass Spectrometry-Based Metabolomics.

Inconsistent peak picking outcomes are a critical concern in processing liquid chromatography-mass spectrometry (LC-MS)-based untargeted metabolomics data. This work systematically studied the mechanisms behind the discrepancies among five commonly used peak picking algorithms, including CentWave in XCMS, linear-weighted moving average in MS-DIAL, automated data analysis pipeline (ADAP) in MZmine 2, Savitzky-Golay in El-MAVEN, and FeatureFinderMetabo in OpenMS. We first collected 10 public metabolomics datasets representing various LC-MS analytical conditions. We then incorporated several novel strategies to (i) acquire the optimal peak picking parameters of each algorithm for a fair comparison, (ii) automatically recognize false metabolic features with poor chromatographic peak shapes, and (iii) evaluate the real metabolic features that are missed by the algorithms. By applying these strategies, we compared the true, false, and undetected metabolic features in each data processing outcome. Our results show that linear-weighted moving average consistently outperforms the other peak picking algorithms. To facilitate a mechanistic understanding of the differences, we proposed six peak attributes: ideal slope, sharpness, peak height, mass deviation, peak width, and scan number. We also developed an R program to automatically measure these attributes for detected and undetected true metabolic features. From the results of the 10 datasets, we concluded that four peak attributes, including ideal slope, scan number, peak width, and mass deviation, are critical for the detectability of a peak. For instance, the focus on ideal slope critically hinders the extraction of true metabolic features with low ideal slope scores in linear-weighted moving average, Savitzky-Golay, and ADAP. The relationships between peak picking algorithms and peak attributes were also visualized in a principal component analysis biplot. Overall, the clear comparison and explanation of the differences between peak picking algorithms can lead to the design of better peak picking strategies in the future.

Pickaxe: a Python library for the prediction of novel metabolic reactions

BackgroundBiochemical reaction prediction tools leverage enzymatic promiscuity rules to generate reaction networks containing novel compounds and reactions. The resulting reaction networks can be used for multiple applications such as designing novel biosynthetic pathways and annotating untargeted metabolomics data. It is vital for these tools to provide a robust, user-friendly method to generate networks for a given application. However, existing tools lack the flexibility to easily generate networks that are tailor-fit for a user’s application due to lack of exhaustive reaction rules, restriction to pre-computed networks, and difficulty in using the software due to lack of documentation.ResultsHere we present Pickaxe, an open-source, flexible software that provides a user-friendly method to generate novel reaction networks. This software iteratively applies reaction rules to a set of metabolites to generate novel reactions. Users can select rules from the prepackaged JN1224min ruleset, derived from MetaCyc, or define their own custom rules. Additionally, filters are provided which allow for the pruning of a network on-the-fly based on compound and reaction properties. The filters include chemical similarity to target molecules, metabolomics, thermodynamics, and reaction feasibility filters. Example applications are given to highlight the capabilities of Pickaxe: the expansion of common biological databases with novel reactions, the generation of industrially useful chemicals from a yeast metabolome database, and the annotation of untargeted metabolomics peaks from an E. coli dataset.ConclusionPickaxe predicts novel metabolic reactions and compounds, which can be used for a variety of applications. This software is open-source and available as part of the MINE Database python package (https://pypi.org/project/minedatabase/) or on GitHub (https://github.com/tyo-nu/MINE-Database). Documentation and examples can be found on Read the Docs (https://mine-database.readthedocs.io/en/latest/). Through its documentation, pre-packaged features, and customizable nature, Pickaxe allows users to generate novel reaction networks tailored to their application.

MAW: the reproducible Metabolome Annotation Workflow for untargeted tandem mass spectrometry

Mapping the chemical space of compounds to chemical structures remains a challenge in metabolomics. Despite the advancements in untargeted liquid chromatography-mass spectrometry (LC–MS) to achieve a high-throughput profile of metabolites from complex biological resources, only a small fraction of these metabolites can be annotated with confidence. Many novel computational methods and tools have been developed to enable chemical structure annotation to known and unknown compounds such as in silico generated spectra and molecular networking. Here, we present an automated and reproducible Metabolome Annotation Workflow (MAW) for untargeted metabolomics data to further facilitate and automate the complex annotation by combining tandem mass spectrometry (MS2) input data pre-processing, spectral and compound database matching with computational classification, and in silico annotation. MAW takes the LC-MS2 spectra as input and generates a list of putative candidates from spectral and compound databases. The databases are integrated via the R package Spectra and the metabolite annotation tool SIRIUS as part of the R segment of the workflow (MAW-R). The final candidate selection is performed using the cheminformatics tool RDKit in the Python segment (MAW-Py). Furthermore, each feature is assigned a chemical structure and can be imported to a chemical structure similarity network. MAW is following the FAIR (Findable, Accessible, Interoperable, Reusable) principles and has been made available as the docker images, maw-r and maw-py. The source code and documentation are available on GitHub (https://github.com/zmahnoor14/MAW). The performance of MAW is evaluated on two case studies. MAW can improve candidate ranking by integrating spectral databases with annotation tools like SIRIUS which contributes to an efficient candidate selection procedure. The results from MAW are also reproducible and traceable, compliant with the FAIR guidelines. Taken together, MAW could greatly facilitate automated metabolite characterization in diverse fields such as clinical metabolomics and natural product discovery.

Regularized adversarial learning for normalization of multi-batch untargeted metabolomics data.

Untargeted metabolomics by mass spectrometry is the method of choice for unbiased analysis of molecules in complex samples of biological, clinical or environmental relevance. The exceptional versatility and sensitivity of modern high-resolution instruments allows profiling of thousands of known and unknown molecules in parallel. Inter-batch differences constitute a common and unresolved problem in untargeted metabolomics, and hinder the analysis of multi-batch studies or the intercomparison of experiments. We present a new method, Regularized Adversarial Learning Preserving Similarity (RALPS), for the normalization of multi-batch untargeted metabolomics data. RALPS builds on deep adversarial learning with a three-term loss function that mitigates batch effects while preserving biological identity, spectral properties and coefficients of variation. Using two large metabolomics datasets, we showcase the superior performance of RALPS as compared with six state-of-the-art methods for batch correction. Further, we demonstrate that RALPS scales well, is robust, deals with missing values and can handle different experimental designs. https://github.com/zamboni-lab/RALPS. Supplementary data are available at Bioinformatics online.

A Metabolomic Signature of Obesity and Risk of Colorectal Cancer: Two Nested Case–Control Studies

Obesity is a leading contributor to colorectal cancer (CRC) risk, but the metabolic mechanisms linking obesity to CRC are not fully understood. We leveraged untargeted metabolomics data from two 1:1 matched, nested case-control studies for CRC, including 223 pairs from the US Prostate, Lung, Colorectal, and Ovarian (PLCO) Cancer Screening Trial and 190 pairs from a prospective Chinese cohort. We explored serum metabolites related to body mass index (BMI), constructed a metabolomic signature of obesity, and examined the association between the signature and CRC risk. In total, 72 of 278 named metabolites were correlated with BMI after multiple testing corrections (p FDR < 0.05). The metabolomic signature was calculated by including 39 metabolites that were independently associated with BMI. There was a linear positive association between the signature and CRC risk in both cohorts (p for linear < 0.05). Per 1-SD increment of the signature was associated with 38% (95% CI: 9-75%) and 28% (95% CI: 2-62%) higher risks of CRC in the US and Chinese cohorts, respectively. In conclusion, we identified a metabolomic signature for obesity and demonstrated the association between the signature and CRC risk. The findings offer new insights into the underlying mechanisms of CRC, which is critical for improved CRC prevention.

Similarities on the mode of action of the terpenoids citral and farnesene in Arabidopsis seedlings involve interactions with DNA binding proteins

The sesquiterpene farnesene and the monoterpene citral are phytotoxic natural compounds characterized by a high similarity in macroscopic effects, suggesting an equal or similar mechanism of action when assayed at IC50 concentration. In the present study, a short-time experiment (24 and 48 h) using an imaging spectrofluorometer allowed us to monitor the in-vivo effects of the two molecules, highlighting that both terpenoids were similarly affecting all PSII parameters, even when the effects of citral were quicker in appearing than those of farnesene. The multivariate, univariate, and pathway analyses, carried out on untargeted-metabolomic data, confirmed a clear separation of the plant metabolome in response to the two treatments, whereas similarity in the affected pathways was observed. The main metabolites affected were amino acids and polyamine, which significantly accumulated in response to both treatments. On the contrary, a reduction in sugar content (i.e. glucose and sucrose) was observed. Finally, the in-silico studies demonstrated a similar mechanism of action for both molecules by interacting with DNA binding proteins, although differences concerning the affinity with the proteins with which they could potentially interact were also highlighted. Despite the similarities in macroscopic effects of these two molecules, the metabolomic and in-silico data suggest that both terpenoids share a similar but not equal mechanism of action and that the similar effects observed on the photosynthetic machinery are more imputable to a side effect of molecules-induced oxidative stress.

DBDIpy: a Python library for processing of untargeted datasets from real-time plasma ionization mass spectrometry.

Plasma ionization is rapidly gaining popularity for mass spectrometry (MS)-based studies of volatiles and aerosols. However, data from plasma ionization are delicate to interpret as competing ionization pathways in the plasma create numerous ion species. There is no tool for detection of adducts and in-source fragments from plasma ionization data yet, which makes data evaluation ambiguous. We developed DBDIpy, a Python library for processing and formal analysis of untargeted, time-sensitive plasma ionization MS datasets. Its core functionality lies in the identification of in-source fragments and identification of rivaling ionization pathways of the same analytes in time-sensitive datasets. It further contains elementary functions for processing of untargeted metabolomics data and interfaces to an established ecosystem for analysis of MS data in Python. DBDIpy is implemented in Python (Version ≥ 3.7) and can be downloaded from PyPI the Python package repository (https://pypi.org/project/DBDIpy) or from GitHub (https://github.com/leopold-weidner/DBDIpy). Supplementary data are available at Bioinformatics online.