Decision tree supported substructure prediction of metabolites from GC-MS profiles

Gas chromatography coupled to mass spectrometry (GC-MS) is one of the most widespread routine technologies applied to the large scale screening and discovery of novel metabolic biomarkers. However, currently the majority of mass spectral tags (MSTs) remains unidentified due to the lack of authenticated pure reference substances required for compound identification by GC-MS. Here, we accessed the information on reference compounds stored in the Golm Metabolome Database (GMD) to apply supervised machine learning approaches to the classification and identification of unidentified MSTs without relying on library searches. Non-annotated MSTs with mass spectral and retention index (RI) information together with data of already identified metabolites and reference substances have been archived in the GMD. Structural feature extraction was applied to sub-divide the metabolite space contained in the GMD and to define the prediction target classes. Decision tree (DT)-based prediction of the most frequent substructures based on mass spectral features and RI information is demonstrated to result in highly sensitive and specific detections of sub-structures contained in the compounds. The underlying set of DTs can be inspected by the user and are made available for batch processing via SOAP (Simple Object Access Protocol)-based web services. The GMD mass spectral library with the integrated DTs is freely accessible for non-commercial use at . All matching and structure search functionalities are available as SOAP-based web services. A XML + HTTP interface, which follows Representational State Transfer (REST) principles, facilitates read-only access to data base entities.     

Construction and application of a mass spectral and retention time index database generated from plant GC/EI-TOF-MS metabolite profiles

The non-supervised construction of a mass spectral and retention time index data base (MS/RI library) from a set of plant metabolic profiles covering major organs of potato (Solanum tuberosum), tobacco (Nicotiana tabaccum), and Arabidopsis thaliana, was demonstrated. Typically 300-500 mass spectral components with a signal to noise ratio > or =75 were obtained from GC/EI-time-of-flight (TOF)-MS metabolite profiles of methoxyaminated and trimethylsilylated extracts. Profiles from non-sample controls contained approximately 100 mass spectral components. A MS/RI library of 6205 mass spectral components was accumulated and applied to automated identification of the model compounds galactonic acid, a primary metabolite, and 3-caffeoylquinic acid, a secondary metabolite. Neither MS nor RI alone were sufficient for unequivocal identification of unknown mass spectral components. However library searches with single bait mass spectra of the respective reference substance allowed clear identification by mass spectral match and RI window. Moreover, the hit lists of mass spectral searches were demonstrated to comprise candidate components of highly similar chemical nature. The search for the model compound galactonic acid allowed identification of gluconic and gulonic acid among the top scoring mass spectral components. Equally successful was the exemplary search for 3-caffeoylquinic acid, which led to the identification of quinic acid and of the positional isomers, 4-caffeoylquinic acid, 5-caffeoylquinic acid among other still non-identified conjugates of caffeic and quinic acid. All identifications were verified by co-analysis of reference substances. Finally we applied hierarchical clustering to a complete set of pair-wise mass spectral comparisons of unknown components and reference substances with known chemical structure. We demonstrated that the resulting clustering tree depicted the chemical nature of the reference substances and that most of the nearest neighbours represented either identical components, as judged by co-elution, or conformational isomers exhibiting differential retention behaviour. Unknown components could be classified automatically by grouping with the respective branches and sub-branches of the clustering tree.     

HPLC Retention time prediction for metabolome analysi

Liquid Chromatography Time-of-Flight Mass Spectrometry (LC-TOF-MS) is widely used for profiling metabolite compounds. LC-TOF-MS is a chemical analysis technique that combines the physical separation capabilities of high-pressure liquid chromatography (HPLC) with the mass analysis capabilities of Time-of-Flight Mass Spectrometry (TOF-MS) which utilizes the difference in the flight time of ions due to difference in the mass-to-charge ratio. Since metabolite compounds have various chemical characteristics, their precise identification is a crucial problem of metabolomics research. Contemporaneously analyzed reference standards are commonly required for mass spectral matching and retention time matching, but there are far fewer reference standards than there are compounds in the organism. We therefore developed a retention time prediction method for HPLC to improve the accuracy of identification of metabolite compounds. This method uses a combination of Support Vector Regression and Multiple Linear Regression adaptively to the measured retention time. We achieved a strong correlation (correlation coefficient = 0.974) between measured and predicted retention times for our experimental data. We also demonstrated a successful identification of an E. coli metabolite compound that cannot be identified by precise mass alone.     

Nearline acquisition and processing of liquid chromatography-tandem mass spectrometry data

Liquid chromatography–mass spectrometry (LC–MS) is a commonly used analytical platform for non-targeted metabolite profiling experiments. Although data acquisition, processing and statistical analyses are almost routine in such experiments, further annotation and subsequent identification of chemical compounds are not. For identification, tandem mass spectra provide valuable information towards the structure of chemical compounds. These are typically acquired online, in data-dependent mode, or offline, using handcrafted acquisition methods and manually extracted from raw data. Here, we present several methods to fast-track and improve both the acquisition and processing of LC–MS/MS data. Our nearly online (nearline) data-dependent tandem MS strategy creates a minimal set of LC–MS/MS acquisition methods for relevant features revealed by a preceding non-targeted profiling experiment. Using different filtering criteria, such as intensity or ion type, the acquisition of irrelevant spectra is minimized. Afterwards, LC–MS/MS raw data are processed with feature detection and grouping algorithms. The extracted tandem mass spectra can be used for both library search and de-novo identification methods. The algorithms are implemented in the R package MetShot and support the export to Bruker, Agilent or Waters QTOF instruments and the vendor-independent TraML standard. We evaluate the performance of our workflow on a Bruker micrOTOF-Q by comparison of automatically acquired and extracted tandem mass spectra obtained from a mixture of natural product standards against manually extracted reference spectra. Using Arabidopsis thaliana wild-type and biosynthetic gene knockout plants, we characterize the metabolic products of a biosynthetic pathway and demonstrate the integration of our approach into a typical non-targeted metabolite profiling workflow.     

Nearline acquisition and processing of liquid chromatography-tandem mass spectrometry data

Liquid chromatography–mass spectrometry (LC–MS) is a commonly used analytical platform for non-targeted metabolite profiling experiments. Although data acquisition, processing and statistical analyses are almost routine in such experiments, further annotation and subsequent identification of chemical compounds are not. For identification, tandem mass spectra provide valuable information towards the structure of chemical compounds. These are typically acquired online, in data-dependent mode, or offline, using handcrafted acquisition methods and manually extracted from raw data. Here, we present several methods to fast-track and improve both the acquisition and processing of LC–MS/MS data. Our nearly online (nearline) data-dependent tandem MS strategy creates a minimal set of LC–MS/MS acquisition methods for relevant features revealed by a preceding non-targeted profiling experiment. Using different filtering criteria, such as intensity or ion type, the acquisition of irrelevant spectra is minimized. Afterwards, LC–MS/MS raw data are processed with feature detection and grouping algorithms. The extracted tandem mass spectra can be used for both library search and de-novo identification methods. The algorithms are implemented in the R package MetShot and support the export to Bruker, Agilent or Waters QTOF instruments and the vendor-independent TraML standard. We evaluate the performance of our workflow on a Bruker micrOTOF-Q by comparison of automatically acquired and extracted tandem mass spectra obtained from a mixture of natural product standards against manually extracted reference spectra. Using Arabidopsis thaliana wild-type and biosynthetic gene knockout plants, we characterize the metabolic products of a biosynthetic pathway and demonstrate the integration of our approach into a typical non-targeted metabolite profiling workflow.

     

Aligning extracted LC-MS peak lists via density maximization

Rapid improvements in mass spectrometry sensitivity and mass accuracy combined with improved liquid chromatography separation technologies allow acquisition of high throughput metabolomics data, providing an excellent opportunity to understand biological processes. While spectral deconvolution software can identify discrete masses and their associated isotopes and adducts, the utility of metabolomic approaches for many statistical analyses such as identifying differentially abundant ions depends heavily on data quality and robustness, especially, the accuracy of aligning features across multiple biological replicates. We have developed a novel algorithm for feature alignment using density maximization. Instead of a greedy iterative, hence local, merging strategy, which has been widely used in the literature and in commercial applications, we apply a global merging strategy to improve alignment quality. Using both simulated and real data, we demonstrate that our new algorithm provides high map (e.g. chromatogram) coverage, which is critically important for non-targeted comparative metabolite profiling of highly replicated biological datasets.

     

Assigning precursor–product ion relationships in indiscriminant MS/MS data from non-targeted metabolite profiling studies

Tandem mass spectrometry using precursor ion selection (MS/MS) is an invaluable tool for structural elucidation of small molecules. In non-targeted metabolite profiling studies, instrument duty cycle limitations and experimental costs have driven efforts towards alternate approaches. Recently, researchers have begun to explore methods for collecting indiscriminant MS/MS (idMS/MS) data in which the fragmentation process does not involve precursor ion isolation. While this approach has many advantages, importantly speed, sensitivity and coverage, confident assignment of precursor–product ion relationships is challenging, which has inhibited broad adoption of the technique. Here, we present an approach that uses open source software to improve the assignment of precursor–product relationships in idMS/MS data by appending a dataset-wide correlational analysis to existing tools. The utility of the approach was demonstrated using a dataset of standard compounds spiked into a malt-barley background, as well as unspiked human serum. The workflow was able to recreate idMS/MS spectra which are highly similar to standard MS/MS spectra of authentic standards, even in the presence of a complex matrix background. The application of this approach has the potential to generate high quality idMS/MS spectra for each detectable molecular feature, which will streamline the identification process for non-targeted metabolite profiling studies.     

Exploring the temperature-stress metabolome of Arabidopsis

Metabolic profiling analyses were performed to determine metabolite temporal dynamics associated with the induction of acquired thermotolerance in response to heat shock and acquired freezing tolerance in response to cold shock. Low-M(r) polar metabolite analyses were performed using gas chromatography-mass spectrometry. Eighty-one identified metabolites and 416 unidentified mass spectral tags, characterized by retention time indices and specific mass fragments, were monitored. Cold shock influenced metabolism far more profoundly than heat shock. The steady-state pool sizes of 143 and 311 metabolites or mass spectral tags were altered in response to heat and cold shock, respectively. Comparison of heat- and cold-shock response patterns revealed that the majority of heat-shock responses were shared with cold-shock responses, a previously unknown relationship. Coordinate increases in the pool sizes of amino acids derived from pyruvate and oxaloacetate, polyamine precursors, and compatible solutes were observed during both heat and cold shock. In addition, many of the metabolites that showed increases in response to both heat and cold shock in this study were previously unlinked with temperature stress. This investigation provides new insight into the mechanisms of plant adaptation to thermal stress at the metabolite level, reveals relationships between heat- and cold-shock responses, and highlights the roles of known signaling molecules and protectants.     

Liquid chromatography-mass spectrometry for comprehensive profiling of ceramide molecules in human hair

Ceramides (CERs) play key roles in signal transduction and cell regulation, probably during the keratinization of human hair. Current methods using mass spectrometry (MS), however, are not sufficient to allow the comprehensive analysis of CER molecules, including isobaric and isomeric CERs. Therefore, a method for the comprehensive profiling of CERs was developed. The method developed is based on reversed-phase liquid chromatography (RPLC) coupled to atmospheric pressure chemical ionization (APCI)-MS. Comprehensive identification and profiling of CERs is achieved using two sets of multimass chromatograms obtained from two channel detections that monitor both molecular-related and sphingoid-related ions under two different in-source collision-induced dissociation conditions and using retention times obtained from RPLC. The application of this method revealed that human hair contains 73 species of CER molecules, which were all corroborated by structural analysis using tandem mass spectrometry. The results further revealed that the composition is characterized by predominant molecules consisting of even carbon atom-containing saturated/unsaturated nonhydroxy or alpha-hydroxy fatty acids and C(18) dihydrosphingosine, a minor but distinct content of isobaric/isomeric and odd chain-containing CERs. This successfully developed RPLC-APCI-MS technique allows the comprehensive profiling of CER molecules in hair for the investigation of their physicochemical and physiological roles.     

Baitmet,a computational approach for GC–MS library-driven metabolite profiling

Introduction

Current computational tools for gas chromatography—mass spectrometry (GC–MS) metabolomics profiling do not focus on metabolite identification, that still remains as the entire workflow bottleneck and it relies on manual data reviewing. Metabolomics advent has fostered the development of public metabolite repositories containing mass spectra and retention indices, two orthogonal properties needed for metabolite identification. Such libraries can be used for library-driven compound profiling of large datasets produced in metabolomics, a complementary approach to current GC–MS non-targeted data analysis solutions that can eventually help to assess metabolite identities more efficiently.

Results

This paper introduces Baitmet, an integrated open-source computational tool written in R enclosing a complete workflow to perform high-throughput library-driven GC–MS profiling in complex samples. Baitmet capabilities were assayed in a metabolomics study involving 182 human serum samples where a set of 61 metabolites were profiled given a reference library.

Conclusions

Baitmet allows high-throughput and wide scope interrogation on the metabolic composition of complex samples analyzed using GC–MS via freely available spectral data. Baitmet is freely available at http://CRAN.R-project.org/package=baitmet.

     

An approach towards method development for untargeted urinary metabolite profiling in metabonomic research using UPLC/QToF MS

The application of LC-MS for untargeted urinary metabolite profiling in metabonomic research has gained much interest in recent years. However, the effects of varying sample pre-treatments and LC conditions on generic metabolite profiling have not been studied. We aimed to evaluate the effects of varying experimental conditions on data acquisition in untargeted urinary metabolite profiling using UPLC/QToF MS. In-house QC sample clustering was used to monitor the performance of the analytical platform. In terms of sample pre-treatment, results showed that untreated filtered urine yielded the highest number of features but dilution with methanol provided a more homogenous urinary metabolic profile with less variation in number of features and feature intensities. An increased cycle time with a lower flow rate (400mul/min vs 600mul/min) also resulted in a higher number of features with less variability. The step elution gradient yielded the highest number of features and the best chromatographic resolution among three different elution gradients tested. The maximum retention time and mass shift were only 0.03min and 0.0015Da respectively over 600 injections. The analytical platform also showed excellent robustness as evident by tight QC sample clustering. To conclude, we have investigated LC conditions by studying variability and repeatability of LC-MS data for untargeted urinary metabolite profiling.     

Aligning extracted LC-MS peak lists via density maximization

Rapid improvements in mass spectrometry sensitivity and mass accuracy combined with improved liquid chromatography separation technologies allow acquisition of high throughput metabolomics data, providing an excellent opportunity to understand biological processes. While spectral deconvolution software can identify discrete masses and their associated isotopes and adducts, the utility of metabolomic approaches for many statistical analyses such as identifying differentially abundant ions depends heavily on data quality and robustness, especially, the accuracy of aligning features across multiple biological replicates. We have developed a novel algorithm for feature alignment using density maximization. Instead of a greedy iterative, hence local, merging strategy, which has been widely used in the literature and in commercial applications, we apply a global merging strategy to improve alignment quality. Using both simulated and real data, we demonstrate that our new algorithm provides high map (e.g. chromatogram) coverage, which is critically important for non-targeted comparative metabolite profiling of highly replicated biological datasets.     

A dynamic programming approach for the alignment of signal peaks in multiple gas chromatography-mass spectrometry experiments

Background  

Gas chromatography-mass spectrometry (GC-MS) is a robust platform for the profiling of certain classes of small molecules in biological samples. When multiple samples are profiled, including replicates of the same sample and/or different sample states, one needs to account for retention time drifts between experiments. This can be achieved either by the alignment of chromatographic profiles prior to peak detection, or by matching signal peaks after they have been extracted from chromatogram data matrices. Automated retention time correction is particularly important in non-targeted profiling studies.     

Identification of metabolites from liquid chromatography–coulometric array detection profiling: Gas chromatography–mass spectrometry and refractionation provide essential information orthogonal to LC–MS/microNMR

Liquid chromatography–coulometric array detection (LC–EC) is a sensitive, quantitative, and robust metabolomics profiling tool that complements the commonly used mass spectrometry (MS) and nuclear magnetic resonance (NMR)-based approaches. However, LC–EC provides little structural information. We recently demonstrated a workflow for the structural characterization of metabolites detected by LC–EC profiling combined with LC–electrospray ionization (ESI)–MS and microNMR. This methodology is now extended to include (i) gas chromatography (GC)–electron ionization (EI)–MS analysis to fill structural gaps left by LC–ESI–MS and NMR and (ii) secondary fractionation of LC-collected fractions containing multiple coeluting analytes. GC–EI–MS spectra have more informative fragment ions that are reproducible for database searches. Secondary fractionation provides enhanced metabolite characterization by reducing spectral overlap in NMR and ion suppression in LC–ESI–MS. The need for these additional methods in the analysis of the broad chemical classes and concentration ranges found in plasma is illustrated with discussion of four specific examples: (i) characterization of compounds for which one or more of the detectors is insensitive (e.g., positional isomers in LC–MS, the direct detection of carboxylic groups and sulfonic groups in ¹H NMR, or nonvolatile species in GC–MS), (ii) detection of labile compounds, (iii) resolution of closely eluting and/or coeluting compounds, and (iv) the capability to harness structural similarities common in many biologically related, LC–EC-detectable compounds.     

Annotating unknown components from GC/EI-MS-based metabolite profiling experiments using GC/APCI(+)-QTOFMS

GC/EI-MS-based metabolite profiling of derivatized polar fractions of crude plant extracts typically reveals several hundred components. Thereof, only up to one half can be identified using mass spectral and retention index libraries, the rest remains unknown. In the present work, the utility of GC/APCI(+)-QTOFMS for the annotation of unknown components was explored. Hence, EI and APCI(+) mass spectra of ~100 known components were extracted from GC/EI-QMS and GC/APCI(+)-QTOFMS profiles obtained from a methoximated and trimethylsilylated root extract of Arabidopsis thaliana. Based on this reference set, adduct and fragment ion formation under APCI(+) conditions was examined and the calculation of elemental compositions evaluated. During these studies, most of the components formed dominating protonated molecular ions. Despite the high mass accuracy (|Δm| ≤ 3 mDa) and isotopic pattern accuracy (mSigma ≤ 30) the determination of a component’s unique native elemental composition requires additional information, namely the number of trimethylsilyl and methoxime moieties as well as the analysis of corresponding collision-induced dissociation (CID) mass spectra. After all, the reference set was used to develop a strategy for the pairwise assignment of EI and APCI(+) mass spectra. Proceeding from these findings, the annotation of unidentified components detected by GC/EI-QMS using GC/APCI(+)-QTOFMS and corresponding deuterated derivatization reagents was attempted. For a total of 25 unknown components, pairs of EI and APCI(+) mass spectra were compiled and elemental compositions determined. Integrative interpretation of EI and CID mass spectra resulted in 14 structural hypotheses, of which seven were confirmed using authenticated standards.