Introducing Undergraduate Students to Metabolomics Using Liquid

Mar 22, 2019 - Introducing Undergraduate Students to Metabolomics Using Liquid Chromatography–High Resolution Mass Spectrometry Analysis of Horse ...
0 downloads 0 Views 2MB Size
Laboratory Experiment Cite This: J. Chem. Educ. XXXX, XXX, XXX−XXX

pubs.acs.org/jchemeduc

Introducing Undergraduate Students to Metabolomics Using Liquid Chromatography−High Resolution Mass Spectrometry Analysis of Horse Blood Mary C. Boyce,†,‡,§ Nathan G. Lawler,†,‡,§ Yingqi Tu,† and Stacey N. Reinke*,†,‡ †

School of Science, Edith Cowan University, Perth 6027, Australia Centre for Integrative Metabolomics and Computational Biology, Edith Cowan University, Perth 6027, Australia



Downloaded via UNIV OF LOUISIANA AT LAFAYETTE on March 22, 2019 at 16:15:26 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

S Supporting Information *

ABSTRACT: Metabolomics is the data-driven science of small molecules. Untargeted metabolomics, using liquid chromatography− high resolution mass spectrometry (LC-HRMS), differs greatly from targeted metabolite assays, using nominal mass LC−MS instruments, as it generates thousands of metabolite features which are not quantified and are identified post hoc. Thus, a substantial amount of time is dedicated to the data processing workflow. Despite the prevalence of untargeted metabolomics and LC−HRMS in contemporary research, undergraduate education in this area is almost nonexistent. To expose upper-division undergraduate analytical chemistry students to untargeted metabolomics, a realistic laboratory experiment, typical of biomedical research, was developed. In the clinic, hemolysis can result from poor sample handling. In this experiment, students artificially induced hemolysis in horse blood and assessed the resulting metabolomic differences. Using XCMS Online, an open-source online platform, and a guided worksheet, students navigated their processed data, learning how untargeted metabolomics differs from the targeted assays they previously performed. This experiment guided their understanding of key concepts such as the number of metabolite features detected, quality assessment, metabolite identification, and data visualization. KEYWORDS: Upper-Division Undergraduate, Laboratory Instruction, Analytical Chemistry, Hands-On Learning/Manipulatives, Mass Spectrometry, Bioanalytical Chemistry



INTRODUCTION Metabolomics is the study of the complement of low molecular weight molecules (metabolites) in a given biological system.1 Contrary to targeted metabolic assays, where a small number of metabolites are assessed to test a hypothesis, metabolomics is a data-driven science, where the aim is to analyze patterns of metabolic changes in order to inform biology and generate hypotheses. Since its introduction, the field has expanded rapidly and is now applied to a variety of contexts, including biomedicine,2 exercise science,3 food and nutrition,4 and plant science.5 The most commonly used analytical platform in metabolomics is liquid chromatography−high resolution mass spectrometry (LC−HRMS). Metabolomics, using LC− HRMS, differs substantially from targeted metabolite assays, using nominal mass instruments (Figure 1). The untargeted approach uses fit-for-purpose analytical methods that are applied to a variety of sample types, and sample preparation generally involves deproteination only. LC−HRMS detects thousands of metabolite “features”, each of which are the m/z of a single ion detected to 3 or 4 decimal places (i.e., exact mass). These metabolite features include adducts, isotopes, and in-source fragmentation products. Thus, one metabolite © XXXX American Chemical Society and Division of Chemical Education, Inc.

may be represented by several metabolite features. Putative identification of these features is based on matching m/z and/ or MS/MS fragmentation patterns to online spectral databases6 or in-house chemical reference libraries.7 As such, a substantial amount of time in the untargeted pipeline is devoted to the computational workflow (data processing, cheminformatics, metabolite identification, statistics, and bioinformatics). In the undergraduate analytical chemistry program at Edith Cowan University, both the theory and application of mass spectrometry constitute an increasing and important part of the curriculum. While the theory of HRMS is taught in lectures, until this activity, the associated laboratory activities focused solely on targeted quantitative methods using nominal mass resolution instruments such as the triple quadruple mass spectrometer. As HRMS is central to metabolomics, designing an experiment around metabolomics provides an excellent application of HRMS. A review of the chemistry education literature highlighted the near absence of metabolomics-based Received: August 3, 2018 Revised: March 3, 2019

A

DOI: 10.1021/acs.jchemed.8b00625 J. Chem. Educ. XXXX, XXX, XXX−XXX

Journal of Chemical Education

Laboratory Experiment

Figure 1. Schematic illustrating differences between targeted LC−MS assays and untargeted LC−HRMS metabolomics studies.

15 min was necessary to induce significant hemolysis. The whole blood was then centrifuged to separate out the plasma. Hemolysis was confirmed by comparison with a colorimetric chart. Plasma samples were deproteinated using ice-cold LC−MS grade methanol. For this experiment an Orbitrap LC−HRMS was employed; however, any LC−HRMS platform such as time of flight (TOF) MS is suitable. While LC−HRMS is becoming increasingly available, it is still relatively expensive and thus may present a barrier for smaller primarily undergraduate institutions. In the absence of access to LC− HRMS facilities, an idealized data set is publicly available in XCMS Online (Job 1222511). A C18 column, with a mobile phase consisting of LC−MS grade acetonitrile, water, and formic acid, was used for this experiment. XCMS Online11 was used for data processing and analysis in this experiment as it is open-source and uses a graphical interface, which makes it user-friendly.12,13 Vendor specific software is expensive and has limitations on the number of users per license, and the XCMS scripts in R can be difficult for beginners to use. Prelab Exercise. Students were asked to read two journal articles: one focusing on hemolysis and its potential impact in metabolomics-based studies,14 and the second providing a background on horse biofluids, including blood metabolites.15 The students then compiled a list of analytes, with associated monoisotopic masses that were expected to be present in horse blood or associated with hemolysis. Sample Preparation. Students worked in groups of 3 or 4 for this part of the experiment. Each group worked with six samples per experimental group. Hemolysis was induced via aspiration with a fine gauge needle, followed by sonication. The blood samples were then prepared for LC−HRMS analysis using standard protocols: plasma was separated from whole blood using centrifugation, and deproteinated using icecold methanol. Quality control (QC) samples were prepared by pooling small aliquots of all plasma samples (nonhemolyzed and hemolyzed). The QC samples were analyzed throughout the LC−HRMS sequence to monitor instrument drift and analytical precision.16 To account for mistakes or lack of hemolysis being achieved in some samples, each student group chose the best four samples per experimental group to analyze. The operation of LC−HRMS instrumentation was demonstrated and carried out by the instructor.

experiments at the undergraduate level. Only one metabolomics laboratory experiment has been published in this Journal, and it used NMR spectroscopy.8 To provide an authentic experience that reflected a real-life metabolomics study, the experiment was designed to use blood plasma. Blood serum and plasma are commonly used in metabolomics studies; however, poor blood collection and handling techniques can lead to hemolysis and therefore affect the metabolome.9,10 The pedagogical aims of this experiment were to enhance the curriculum by including high resolution mass spectrometry in the laboratory; introduce final year students to untargeted metabolomics, a rapidly growing field that has a strong analytical chemistry component; and broaden students’ understanding of the data processing and quality metric differences between untargeted and targeted assays. Herein, a metabolomics laboratory experiment that investigates the effect of hemolysis on the horse plasma metabolome, using LC− HRMS, is presented.



EXPERIMENTAL DETAILS This experiment is delivered in a third (final) year analytical chemistry course which comprises a weekly 2 h lecture and 3 h laboratory for a duration of 13 weeks (one semester). The course focuses (approximately 80%) on separation techniques and associated detection methods, one of which is MS. The theory of MS, including high resolution instruments, is covered in lectures. This experiment runs in the final weeks of the semester, and prior to it, students complete several targeted assays using both LC and gas chromatography (GC) with a number of detectors including MS (e.g., determination of short chain fatty acids using GC−MS). The use of internal standards and preparation of calibration standards are routinely applied in these assays. The enrollment number for this subject is typically 30−40, so the students are split into two 3 h laboratory classes. Two instructors deliver each laboratory session. Full details of the experimental procedure are provided in the Supporting Information (see files detailing experiment supplied to students and notes for instructor). Week 1: Hemolysis and Sample Preparation

Laboratory Requirements. In this experiment, a single sample of commercially purchased horse whole blood was used; however, any whole blood sample containing an anticoagulant would be appropriate. The combination of using a fine gauge needle (25 gauge) followed by sonication for B

DOI: 10.1021/acs.jchemed.8b00625 J. Chem. Educ. XXXX, XXX, XXX−XXX

Journal of Chemical Education

Laboratory Experiment

Week 2: Data Processing and Interpretation of Results

Prelab Exercise. Students were given access to their LC− HRMS raw files several days before the computer laboratory session and asked to upload them onto XCMS Online.11 XCMS Online is an open-source mass spectral data processing platform which performs feature detection, retention time alignment, putative annotation, statistics, and data visualization; in-depth descriptions of its functionality have been previously published.12,13 Following the instructional guideline (Supporting Information file detailing experiment supplied to students, pages S6−S10), the students set the processing parameters and processed their data. Interpretation of Results. During the computer laboratory session, the instructor gave a short prelab presentation on untargeted data processing as well as a live-action demonstration of the various ways to explore results on the XCMS Online platform. Working in pairs, the students were provided a worksheet (Supporting Information, computer lab worksheet), with guided tasks and questions that were designed to facilitate students’ understanding of the differences between targeted MS assays that they have previously performed and untargeted LC−HRMS-based metabolomics. To achieve this pedagogical aim, the questions focused on the number of features detected, assessment of the quality of peaks, and annotation (putative identification) of features. Although statistical analysis of metabolomics data was beyond the scope of this experiment and the primary learning objectives, the students did visualize the outcome of their experiment with the Principal Components Analysis (PCA)17,18 scores plot generated by XCMS Online.

Figure 2. Levels of hemolysis achieved. Plasma was separated from whole blood, using centrifugation, following artificial induction of hemolysis. The relative level of hemolysis was easily identifiable for students. The plasma of samples either not subjected to artificial hemolysis or not successfully hemolyzed (far left) maintained a yellow color. Successfully hemolyzed samples resulted in varying levels of red-colored plasma (center and right).

colorimetric chart (in the experiment supplied to students) was used to record the level of hemolysis. The procedure of deproteination using methanol also served as an abbreviated learning opportunity about protein tertiary structure and denaturation/precipitation using organic solvents. In the computational component of the experiment, students used the XCMS Online12,13 platform to process their data files and interpret the results. The intricacies of how XCMS processes the data were not important; however, a general overview was provided so that students were aware of how the data were processed. XCMS Online provides several options for interpreting and visualizing results. The results table (Figure 3) provides a list of all detected features. Where possible, metabolite annotations (putative identifications) are made by screening the m/z against the Metlin database.19 XCMS Online also provides images to visualize retention time correction as well as a PCA scores plot, generated from all detected features. A worksheet was used to guide students through various features of XCMS Online, directing their learning and conceptualization of the differences between targeted LC−MS assays and untargeted LC−HRMS metabolomics. The students were first directed to the results table, which listed over 10,000 aligned metabolite features and served to highlight the enormity of the data challenge in metabolomics. To help students engage with the data and begin to understand the analytical and quality control differences between targeted assays and metabolomics, they were asked to assess the EICs and box plots of different features. As analytical chemists, the students were used to seeing Gaussian chromatographic peaks in their targeted assays. By visualizing various EICs from the untargeted analysis, the students realized that not all peaks were Gaussian and some were actually integrated from the noise. The QC samples were included in the data processing and, for the purposes of this experiment, considered a third experimental group. In typical untargeted metabolomics studies, the QCs are used to monitor analytical drift and measure precision.16 Features not meeting acceptable precision thresholds (example: >20% relative standard deviation (RSD)) are removed from the data set prior to statistical analysis. While RSD calculation and data cleaning were beyond the scope of this experiment, the students qualitatively assessed the reproducibility of the QC samples using the box plots; a tightly dispersed box indicated low variance in the QC samples and



HAZARDS Horse blood is a potentially infectious substance. Students were required to work in the fume hood with disposable gloves and safety glasses. All material in contact with the blood was placed in an autoclave bag for disposal. Centrifuges with a rubber seal were used in the event of leakage from the centrifuge tubes. The fume hood was sprayed with 70% ethanol at the conclusion of the experiment. The experiment required students to use a fine needle to shear/rupture the blood cells, so students were at risk of receiving a needle stick injury. Therefore, students were instructed on the safe use and disposal of needles. Methanol is a hazardous organic solvent. It is toxic if inhaled, ingested, or absorbed through the skin, and it can cause damage to organs. LC mobile phases were prepared by the instructor. Users should consult the material safety data sheets for a chemical prior to its use.



RESULTS AND DISCUSSION This experiment was trialled and subsequently optimized with three individual undergraduate students prior to its transfer to the classroom. It was then successfully carried out by two groups, each containing 16 third year students. In the laboratory component of the experiment (week 1), the students induced hemolysis in horse blood, separated the plasma component from the blood cells, and then deproteinized the plasma in preparation for LC−HRMS analysis. Using the preoptimized methodology of aspiration through a fine gauge needle followed by sonication, all students were successfully able to induce hemolysis in at least 80% of their samples. The extent of hemolysis did vary (Figure 2), and a C

DOI: 10.1021/acs.jchemed.8b00625 J. Chem. Educ. XXXX, XXX, XXX−XXX

Journal of Chemical Education

Laboratory Experiment

Figure 3. Abbreviated results table and images created in XCMS Online. The results table provides peak information, including the following: arbitrary feature number (featureidx), p-value from univariate statistical test (pvalue), median m/z and retention time for each feature (mzmed and rtmed, respectively), possible adducts (adducts), and peak grouping (peakgroup). In this image, features 1, 2, and 4 are assigned to the same peak group (#137) and likely belong to the same metabolite. When specific rows are clicked and highlighted, corresponding extracted ion chromatograms (EIC), mass spectra, box plots, and annotations appear on the screen (Supporting Information, notes for instructors, p S3).

Figure 4. EICs and box plots of example peaks. (A) Peak 583.2540 m/z at 10.56 min (not annotated) had a Gaussian shape but low analytical precision. On the basis of peak shape alone, the students considered this to be a good feature; however, the box plot shows that the peak may not be reproducibly detected. (B) Peak 586.2722 m/z at 5.78 min (annotated as a 4 amino acid peptide) had a non-Gaussian shape but a high analytical precision. On the basis of peak shape, the students considered this to be an undesirable feature for data analysis; however, the QC signal shows excellent reproducibility in the box plot.

therefore high analytical precision. Figure 4 illustrates that Gaussian peaks can be associated with poor analytical precision

and, equally, non-Gaussian peaks can exhibit a high degree of precision. Thus, the students learned that features associated D

DOI: 10.1021/acs.jchemed.8b00625 J. Chem. Educ. XXXX, XXX, XXX−XXX

Journal of Chemical Education

Laboratory Experiment

with “poor” chromatographic peaks can still be included in metabolomics data analysis, if they are reproducibly detected. The focus of the activity then moved to metabolite identification and biological results. The students were asked to record the number of possible identifications (based on m/ z) listed for some of the features at which they had been looking. The list of possible identifications often exceeded 10 or even 20, highlighting that m/z alone is insufficient to confirm metabolite identity and that identification can be a challenging task. Several students then inquired about how metabolite identification is achieved in research, which led to an in-class discussion and added learning opportunity about retention time and MS/MS fragmentation pattern matching, and levels of identification confidence.20 With the limitations of analyte identification/annotation in mind, the students tried to find peaks associated with metabolites found in their prelaboratory readings. They found two lysophosphatidylcholines (LysoPC(16:0) and LysoPC(18:0)) reported to increase with hemolysis;14 however, the students’ results showed the opposite trend. The students were also asked to indicate what types of metabolites changed as a result of hemolysis. Different approaches were taken by students to relate the results table back to the biological question. Several students looked for familiar metabolite names in the most significantly different features (i.e., the top listed features in the results table sorted by p-value). These students found a feature that decreased with hemolysis and annotated to bilirubin, a breakdown product of heme. Other students methodically Google-searched annotated names of significant features. One student found a significant feature, annotated as gallamine, a muscle relaxant often used by veterinarians. While the students’ findings and interpretations varied, in terms of significant features and corresponding biological relevance, this exercise helped them to not only understand the general concept of metabolomics as a datadriven, but also provided them some biological meaning to their data. Having interrogated the results table, containing several thousand features, the students quickly appreciated the difficulty in interpreting such a large amount of data. Using a multivariate modeling approach (PCA) of all detected features, it was clear that the metabolome was different between hemolyzed and nonhemolyzed samples (Figure 5). Along principal component (PC) 1 (the direction of maximum variance), there was separation between the three groups, with the greatest degree of intragroup variance being in the hemolysis group, likely due to the varied level of hemolysis obtained. This experiment was an end-of-semester activity that was as much about gaining a general understanding of LC−HRMS and untargeted analysis as it was about exposing students to complex, multidisciplinary, and advanced technologies. Therefore, a less formal postlab worksheet-style assessment was completed, instead of a full scientific report (Supporting Information, assessment exercise), with moderate success. The first question asked students to determine if analytes known to be present in horse blood and also altered with hemolysis (based on their prelab readings) were detected in their samples. Students had to use monoisotopic mass and have an understanding of adduct formation [M + H]+ to successfully search their data and complete this question, providing evidence of their understanding of these key concepts. This question was successfully completed by almost all students. The second question asked students to find features (analytes)

Figure 5. PCA plot showing metabolomic differences of a representative sample set from one student. Data were logtransformed, and scaled to unit variance, and missing values were imputed using k-nearest neighbors. Blue diamonds, nonhemolyzed; red circles, QCs; yellow triangles, hemolyzed samples.

that were significantly affected by hemolysis and interpret the findings, on the basis of literature reports. This question was too open-ended for most students, and they struggled to find examples in the literature that provided biological relevance at a level that was comprehensible to an undergraduate audience. The final question asked students to summarize an introduction of the field of metabolomics. It was evident that most students did obtain a broad understanding of metabolomics and its utility as a field of science; however, students with either a biological science or statistical background demonstrated a more in-depth understanding of the data and how to interpret it. The final examination included questions on how HRMS data collection and interpretation differed from nominal mass instrumentation. It was evident that this experiment helped to provide context for the HRMS lectures and an improved understanding of the content. To further strengthen this understanding, future iterations of the computer laboratory session will be structured to include reflective pause point activities to better connect the data with both biological outcome and differences observed in the HRMS approach. This experiment was challenging, both from the instructors’ perspective in designing an activity that is introductory, but also authentic, and from the students’ perspective in needing biological knowledge, analytical chemistry skills, and new data analysis skills. Most students embraced the activity and, once they were familiar with navigating XCMS Online, were keen to explore the data and make biological connections. Such findings by the students made the data more meaningful and generated interest. From an instructor’s point of view, it was rewarding to inform teaching from research, providing students access to state-of-the-art instrumentation and a contemporary field of science.



CONCLUSION This experiment was primarily designed to broaden students’ exposure to untargeted analysis and the rapidly expanding field of metabolomics. The workflow design highlighted the conceptual differences between targeted LC−MS assays and E

DOI: 10.1021/acs.jchemed.8b00625 J. Chem. Educ. XXXX, XXX, XXX−XXX

Journal of Chemical Education

Laboratory Experiment

(9) Kirwan, J. A.; Brennan, L.; Broadhurst, D.; Fiehn, O.; Cascante, M.; Dunn, W. B.; Schmidt, M. A.; Velagapudi, V. Preanalytical Processing and Biobanking Procedures of Biological Samples for Metabolomics Research: A White Paper, Community Perspective (for ″Precision Medicine and Pharmacometabolomics Task Group″-The Metabolomics Society Initiative). Clin. Chem. 2018, 64 (8), 1158− 1182. (10) Denihan, N. M.; Walsh, B. H.; Reinke, S. N.; Sykes, B. D.; Mandal, R.; Wishart, D. S.; Broadhurst, D. I.; Boylan, G. B.; Murray, D. M. The effect of haemolysis on the metabolomic profile of umbilical cord blood. Clin. Biochem. 2015, 48 (7−8), 534−537. (11) XCMSOnline. https://xcmsonline.scripps.edu (accessed March 2019). (12) Tautenhahn, R.; Patti, G. J.; Rinehart, D.; Siuzdak, G. XCMS Online: a web-based platform to process untargeted metabolomic data. Anal. Chem. 2012, 84 (11), 5035−5039. (13) Gowda, H.; Ivanisevic, J.; Johnson, C. H.; Kurczy, M. E.; Benton, H. P.; Rinehart, D.; Nguyen, T.; Ray, J.; Kuehl, J.; Arevalo, B.; Westenskow, P. D.; Wang, J.; Arkin, A. P.; Deutschbauer, A. M.; Patti, G. J.; Siuzdak, G. Interactive XCMS Online: simplifying advanced metabolomic data processing and subsequent statistical analyses. Anal. Chem. 2014, 86 (14), 6931−6939. (14) Yin, P.; Peter, A.; Franken, H.; Zhao, X.; Neukamm, S. S.; Rosenbaum, L.; Lucio, M.; Zell, A.; Haring, H. U.; Xu, G.; Lehmann, R. Preanalytical aspects and sample quality assessment in metabolomics studies of human blood. Clin. Chem. 2013, 59 (5), 833−845. (15) Escalona, E. E.; Leng, J.; Dona, A. C.; Merrifield, C. A.; Holmes, E.; Proudman, C. J.; Swann, J. R. Dominant components of the Thoroughbred metabolome characterised by (1) H-nuclear magnetic resonance spectroscopy: A metabolite atlas of common biofluids. Equine Vet J. 2015, 47 (6), 721−730. (16) Broadhurst, D.; Goodacre, R.; Reinke, S. N.; Kuligowski, J.; Wilson, I. D.; Lewis, M. R.; Dunn, W. B. Guidelines and considerations for the use of system suitability and quality control samples in mass spectrometry assays applied in untargeted clinical metabolomic studies. Metabolomics 2018, 14 (6), 72. (17) Jolliffe, I. T. Principal Component Analysis; Springer-Verlag: New York, 1986. (18) Bro, R.; Smilde, A. K. Principal component analysis. Anal. Methods 2014, 6 (9), 2812−2831. (19) Benton, H. P.; Ivanisevic, J.; Mahieu, N. G.; Kurczy, M. E.; Johnson, C. H.; Franco, L.; Rinehart, D.; Valentine, E.; Gowda, H.; Ubhi, B. K.; Tautenhahn, R.; Gieschen, A.; Fields, M. W.; Patti, G. J.; Siuzdak, G. Autonomous metabolomics for rapid metabolite identification in global profiling. Anal. Chem. 2015, 87 (2), 884−891. (20) Sumner, L. W.; Amberg, A.; Barrett, D.; Beale, M. H.; Beger, R.; Daykin, C. A.; Fan, T. W.; Fiehn, O.; Goodacre, R.; Griffin, J. L.; Hankemeier, T.; Hardy, N.; Harnly, J.; Higashi, R.; Kopka, J.; Lane, A. N.; Lindon, J. C.; Marriott, P.; Nicholls, A. W.; Reily, M. D.; Thaden, J. J.; Viant, M. R. Proposed minimum reporting standards for chemical analysis Chemical Analysis Working Group (CAWG) Metabolomics Standards Initiative (MSI). Metabolomics 2007, 3 (3), 211−221.

untargeted metabolomics using LC−HRMS, including analytical methods, data processing, quality control, and metabolite identification. It also provided the students with an understanding of the multidisciplinary nature of the metabolomics, such as analytical chemistry, cheminformatics, chemometrics, and biology/applied sciences, as well as data interpretation and visualization. Finally, the experiment served to expose final year students to a more complex, but realistic, experiment, typical of research, just at a time when they are deciding their career path.



ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available on the ACS Publications website at DOI: 10.1021/acs.jchemed.8b00625. Experiment provided to students (PDF, DOCX) Notes for instructors (PDF, DOCX) Computer lab worksheet (PDF, DOCX) Assessment exercise (PDF, DOCX)



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. ORCID

Nathan G. Lawler: 0000-0001-9649-425X Stacey N. Reinke: 0000-0002-0758-0330 Author Contributions §

M.C.B. and N.G.L. contributed equally to this work.

Notes

The authors declare no competing financial interest.



REFERENCES

(1) Fiehn, O. Metabolomics−the link between genotypes and phenotypes. Plant Mol. Biol. 2002, 48 (1−2), 155−171. (2) Beger, R. D.; Dunn, W.; Schmidt, M. A.; Gross, S. S.; Kirwan, J. A.; Cascante, M.; Brennan, L.; Wishart, D. S.; Oresic, M.; Hankemeier, T.; Broadhurst, D. I.; Lane, A. N.; Suhre, K.; Kastenmüller, G.; Sumner, S. J.; Thiele, I.; Fiehn, O.; KaddurahDaouk, R. Metabolomics enables precision medicine: “A White Paper, Community Perspective. Metabolomics 2016, 12 (9), 149. (3) Heaney, L. M.; Deighton, K.; Suzuki, T. Non-targeted metabolomics in sport and exercise science. J. Sports Sci. [Online] 2017, 1−9. (4) Brennan, L.; Hu, F. B. Metabolomics-Based Dietary Biomarkers in Nutritional Epidemiology-Current Status and Future Opportunities. Mol. Nutr. Food Res. 2019, 63, 1701064. (5) Kumar, R.; Bohra, A.; Pandey, A. K.; Pandey, M. K.; Kumar, A. Metabolomics for Plant Improvement: Status and Prospects. Front. Plant Sci. 2017, 8, 1302. (6) Kind, T.; Tsugawa, H.; Cajka, T.; Ma, Y.; Lai, Z.; Mehta, S. S.; Wohlgemuth, G.; Barupal, D. K.; Showalter, M. R.; Arita, M.; Fiehn, O. Identification of small molecules using accurate mass MS/MS search. Mass Spectrom. Rev. 2018, 37 (4), 513−532. (7) Naz, S.; Gallart-Ayala, H.; Reinke, S. N.; Mathon, C.; Blankley, R.; Chaleckis, R.; Wheelock, C. E. Development of a Liquid Chromatography-High Resolution Mass Spectrometry Metabolomics Method with High Specificity for Metabolite Identification Using All Ion Fragmentation Acquisition. Anal. Chem. 2017, 89 (15), 7933− 7942. (8) Sandusky, P. O. Introducing Undergraduate Students to Metabolomics Using a NMR-Based Analysis of Coffee Beans. J. Chem. Educ. 2017, 94 (9), 1324−1328. F

DOI: 10.1021/acs.jchemed.8b00625 J. Chem. Educ. XXXX, XXX, XXX−XXX