1 s2.0 S0003267021008692 Main

Analytica Chimica Acta xxx (xxxx) xxx
Contents lists available at ScienceDirect
Analytica Chimica Acta

journal homepage: www.elsevier.com/locate/aca
Tutorial
Low and high resolution gas chromatography-mass spectrometry for

untargeted metabolomics: A tutorial
Fernanda Rey-Stolle a, Danuta Dudzik a, b, Carolina Gonzalez-Riano a,
ndez-García a, Vanesa Alonso-Herranz a, David Rojo a, Coral Barbas a,
Miguel Ferna
Antonia García a, *
a
Centre for Metabolomics and Bioanalysis (CEMBIO), Faculty of Pharmacy, Universidad San Pablo CEU, CEU Universities. Campus Monteprincipe, Boadilla
Del Monte, 28668, Madrid, Spain
b
Department of Biopharmaceutics and Pharmacodynamics, Faculty of Pharmacy, Medical University of Gdan sk, Poland
h i g h l i g h t s g r a p h i c a l a b s t r a c t
Ready-to-use protocols for untar-

geted GC-MS-based metabolomics
are provided.
Detailed sample treatment proced-
ures for more than 10 types of sam-
ples: biofluids, tissues and cells are
included.
Low and high resolution approaches
(GC-Q-MS vs GC-QTOF-MS) in
metabolomics are discussed.
A case study is included, with
strengths and limitations of untar-
geted metabolomics based on low
and high resolution systems.
a r t i c l e i n f o a b s t r a c t
Article history: GC-MS for untargeted metabolomics is a well-established technique. Small molecules and molecules
Received 29 April 2021 made volatile by derivatization can be measured and those compounds are key players in main biological
Received in revised form pathways. This tutorial provides ready-to-use protocols for GC-MS-based metabolomics, using either the
4 September 2021
well-known low-resolution approach (GC-Q-MS) with nominal mass or the more recent high-resolution
Accepted 6 September 2021
Available online xxx
approach (GC-QTOF-MS) with accurate mass, discussing their corresponding strengths and limitations.
Analytical procedures are covered for different types of biofluids (plasma/serum, bronchoalveolar lavage,
urine, amniotic fluid) tissue samples (brain/hippocampus, optic nerve, lung, kidney, liver, pancreas) and
Keywords:
GC-MS protocols
samples obtained from cell cultures (adipocytes, macrophages, Leishmania promastigotes, mitochondria,
High-resolution mass spectrometry culture media). Together with the sample preparation and data acquisition, data processing strategies are
Metabolic fingerprinting described specially focused on Agilent equipments, including deconvolution software and database
Compound identification annotation using spectral libraries. Manual curation strategies and quality control are also deemed.
Metabolomics or metabonomics Finally, considerations to obtain a semiquantitative value for the metabolites are also described. As a case
Spectral library study, an illustrative example from one of our experiments at CEMBIO Research Centre, is described and
findings discussed.
© 2021 Elsevier B.V. All rights reserved.
* Corresponding author.
E-mail address: antogar@ceu.es (A. García).
https://doi.org/10.1016/j.aca.2021.339043
0003-2670/© 2021 Elsevier B.V. All rights reserved.
Please cite this article as: F. Rey-Stolle, D. Dudzik, C. Gonzalez-Riano et al., Low and high resolution gas chromatography-mass spectrometry for
untargeted metabolomics: A tutorial, Analytica Chimica Acta, https://doi.org/10.1016/j.aca.2021.339043
F. Rey-Stolle, D. Dudzik, C. Gonzalez-Riano et al. Analytica Chimica Acta xxx (xxxx) xxx
Contents
1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
2. Experimental design, preanalytical considerations and quality assurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
2.1. Experimental design and ethical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
2.2. Sample and preanalytical considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
2.3. Quality assurance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
3. Sample treatment before GC-MS analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
3.1. Detailed sample treatment protocols for biofluids . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
3.2. Detailed sample treatment protocols for tissues [29e33] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
3.3. Detailed sample treatment protocols for eukaryotic cells and mitochondria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
3.4. Detailed sample treatment protocols for culture media [42] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
4. Analytical method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
4.1. Instrument set-up . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
4.2. Sequence design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
4.3. Calibration of the instrument . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
4.4. Checklist . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
4.5. Troubleshooting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
5. Data extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
5.1. Quality assurance and samples for quality control . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
5.2. Preprocessing data: deconvolution and searching against spectral libraries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
6. Relative quantification of components and unknown compounds with low and high resolution MS detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
6.1. Data normalization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
6.2. Data filtering . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
6.3. Transformation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
6.4. Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
7. Metabolite Identification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
7.1. Curation of tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
7.2. Potential and experimentally confirmed capabilities of the accurate mass detectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
7.3. Case study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
8. Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Author contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Declaration of competing interest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Acknowledgment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Supplementary data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 00
Abbreviations GC-QTOF-MS Gas Chromatography-Time-of-Flight Mass

Spectrometry
AMDIS Automated Mass Spectral Deconvolution and IS Internal Standard
Identification System MCR Multivariate Curve Resolution
BSTFA N,O-Bis(trimethylsilyl)trifluoroacetamide MS/MS Tandem Mass Spectrometry
BSTFA with 1% TMCS N,O-bis(trimethylsilyl)trifluoroacetamide MSTFA N-Methyl-N-(trimethylsilyl)trifluoroacetamide
with 1% trimethylchlorosilane MTBE tert-butyl methyl ether
CVD Cardiovascular disease MTBSTFA N-tert-butyldimethylsilyl-N-
EI Electron Ionization methyltrifluoroacetamide
EIC Extracted Ion Chromatogram PCDL Personal Compound Database and Library
FAME Fatty Acid Methyl Ester QA Quality Assurance
GC-(EI)-Q-MS Gas Chromatography-Electron Ionization- QC Quality Control
Quadrupole Mass Spectrometry RAMSY Ratio Analysis of Mass Spectrometry
GC-(EI)-QTOF-MS Gas Chromatography-Electron Ionization- RSD Relative Standard Deviation
Time-of-Flight Mass Spectrometry RTL Retention Time Locking
GC-MS Gas chromatography-Mass Spectrometry SNR Signal-to-Noise Ratio
GC-Q-MS Gas Chromatography-Quadrupole Mass TCA Tricarboxylic acid
Spectrometry TIC Total Ion Chromatogram
VOCs Volatile Organic Compounds
1. Introduction compounds, typically <1500 Da) in a given sample obtained from a

biological system [1,2]. Untargeted (or non-targeted) metabolomics
Metabolomics refers to the systematic analysis of the complete encompasses the most comprehensive analysis of metabolites in a
set of intra- and extracellular metabolites (low molecular weight biological sample. The whole metabolomics study involves several
2
Table 1
Comparison of low- and high resolution GC-MS equipments.
FEATURES GC-Q-MS GC-QTOF-MS
Metabolomic approach (targeted, untargeted): absolute and Yes Yes

semi-quantification
Use of the instrument Easy Relatively complex
Chromatographic Resolution High Very high
Robustness Very high Medium
Data processing (time/sample) Seconds Minutes
Computer processing and memory size requirements Medium Very high
Spectral libraries Unit resolution þ Retention time Personal Compound Database and Library (PCDL) with exact
mass þ Retention time
Structural elucidation capabilities for novel compounds Low High (MS/MS)
Specificity High Very high
Purchase Price Medium High
steps, from the design of the experiment to the interpretation and using GC systems coupled to low- and high resolution mass spec-
validation of the obtained results with new experiments [3]. Due to trometry analyzers. Readers will be oriented through the overall
the high diversity of physicochemical properties of metabolites and workflow dedicated to metabolic fingerprinting applications based
the different concentrations at which these are commonly found, a on GC-MS analytical platforms including biological considerations,
combination of different analytical techniques is always recom- study design, and sample treatment for the study of small metab-
mended and among them, gas chromatography coupled to mass olites in biological samples. Besides, method parameters for GC-Q-
spectrometry (GC-MS) is usually included. Before the analysis, MS and GC-QTOF-MS, data acquisition and data processing
sample treatment is required for removing large biomolecules, including deconvolution, compound search against spectral li-
mainly proteins, and converting metabolites into their volatile braries (including NIST library) and compound annotation will be
derivatives through chemical derivatization. For over more than 60 detailed specifically for GC-MS Agilent Technologies systems, by
years, GC-MS standardized protocols have been a fundamental the use of Agilent software tools such as MassHunter Qualitative
methodology for the study of key metabolic pathways such as the and Quantitative Analysis and Unknowns Analysis including Sure-
tricarboxylic acid (TCA) cycle, glycolysis, pentose phosphate Mass deconvolution procedures. This tutorial covers both the
pathway, among others, mainly by the comparison of the peak characterization and the semi-quantitation of metabolites by GC-
abundances of alcohols, sugars, amino acids, amines, amides and MS.
fatty acids [4,5]. Briefly, the main novelties of this tutorial are:
The most representative advantages of this chromatographic
technique are its excellent efficiency, chromatographic resolution, 1. It includes ready-to-use protocols for different type of biological
and high repeatability, both in retention time and response. samples: biofluids (plasma/serum, bronchoalveolar lavage,
Regarding mass spectrometric detection through the most com- urine, amniotic fluid) tissue samples (brain/hippocampus, optic
mon ion source, electron ionization (EI), the possibility of searching nerve, lung, kidney, liver, pancreas) and samples obtained from
dual information (retention time and hard-fragmentation spec- cell cultures (adipocytes, macrophages, Leishmania promasti-
trum) by comparison against commercial and in-house spectral gotes, mitochondria, and culture media). Besides, a novel IS (not
libraries is one of the most valuable attributes of GC-MS. This su- described before) is recommended (d31-palmitic acid (d31-
perior characteristic reaches the top with high-resolution detectors C16:0)) has been chosen because of its best performance and
(time-of-flight, orbitrap) which allow for the determination of ac- price among several candidates previously assayed.
curate mass in all detected m/z (mass-to-charge) fragments. Table 1 2. It encompasses the whole analytical method for sample analysis
depicts the main advantages and limitations of GC-Q-MS (low and data preprocessing with high resolution GC-MS equipment.
resolution) and GC-QTOF-MS (high resolution) instruments. Even In addition, the total ion chromatogram obtained with human
though the former is the most extended for bioanalysis and it can plasma sample and the list of peak assignations are given.
operate continuously provided that a good previous maintenance is 3. It adds the performance comparison of both high-resolution and
performed, unknown compound identification is challenging in low-resolution equipments in a case study.
GC-Q-MS [6]. In order to overcome this limitation, the latter offers 4. It gives recommendations for low-energy sources.
additional tools for the identification of novel compounds based on
accurate mass determination, which enables the application of The obtained information could be applied in the study of
high-quality algorithms facilitating isotope distribution-based different clinical conditions including diabetes, cancer, obesity,
molecular formula determination of molecular ions, spectral simi- cardiovascular disease (CVD), infectious diseases, and renal dis-
larity searches and de novo structural elucidation. Both systems can eases. Stool, as well as plant samples will be excluded, as they
be used for characterization of the metabolic profile, semi- would deserve independent full manuscripts. Special strategies for
quantitative purposes or absolute quantitation of volatile de- the analysis of solely volatile organic compounds VOCs also will be
rivatives of metabolites in any kind of sample or condition. out of the scope of this tutorial.
In a previous tutorial [7], we proposed protocols and procedures This is a straightforward guide that: (i) describes each step of the
for untargeted metabolomics based in low resolution GC-Q-MS metabolic profiling workflow for GC-MS metabolomics analysis for
systems and AMDIS software for deconvolution and library search different types of biological samples, (ii) presents easy-to-use
against target libraries (retention time/retention index and spec- procedures for different biological samples to be performed by
trum). Currently, high-resolution equipments have gained atten- low- or high-resolution GC-MS, (iii) introduces quality issues and
tion due to their more powerful capabilities for unveiling unknown critical discussion of the workflow steps, and (iv) comments on pros
compounds [4,8,9]. The present tutorial focuses on the description, and cons of the high-resolution MS detection.
as well as tips and tricks of untargeted metabolomics workflows
3
2. Experimental design, preanalytical considerations and recommend performing a pilot study with a reduced number of
quality assurance samples ensuring reliable testing the normal distribution of resi-
dues in order to estimate the parameters needed for calculating the
2.1. Experimental design and ethical considerations sample size that allows a power of 0.8. In the simplest case of a
pairwise, two-tailed parametric test variance and distribution of
Important decisions regarding this issue must be considered specific metabolites within each sample group can be estimated
such as number and type of groups, individuals (including inclusion allowing power calculation by available tools such as Metab-
and exclusion criteria), type of samples, associated metadata oAnalyst 5.0 [14].
including confounding variables (e.g., medication, smokers, sample
collection date), growing conditions for cells, as well as the amount 2.2. Sample and preanalytical considerations
of each sample, collection, storage, analytical instruments, and
methodology. A process flow diagram describing the main points to Collection of sample types extensively used in metabolomics
be considered in the experimental design is depicted in Fig. 1. When such as urine is minimally invasive, whereas blood plasma/serum/
considering sample size, studies performed on the human popu- tissue collection implies some kind of injury. Besides, tissue har-
lation entails greater complexity due to the significant biological vesting possesses several challenges due to the highly heteroge-
variation due to factors including age, gender, BMI, lifestyle, diet neous nature. Immediately after sample withdrawal, the
observed in the general population, while animal models are per- metabolism should be quenched to prevent further biochemical
formed under a controlled laboratory environment. Achieving reactions that can modify the metabolic content and produce bias.
ethical approval and, for studies with humans, written informed Quenching methods involve the use of liquid nitrogen and/or
consent from all the participants is mandatory. A comprehensive addition of organic solvents [15]. Storage conditions under low
description of this topic is beyond the scope of this tutorial. Further temperatures, preferably 80 C to guarantee the sample stability.
information can be found elsewhere [10e13]. However, this stability is metabolite species-dependent. The in-
Special considerations regarding the statistical power of untar- fluence of long storage times and sample integrity in preserving
geted metabolomics experiments must be considered. While stra- adequate metabolite concentrations has been investigated [16e18].
tegies aiming to reduce type I errors are well established and Thus, multiple freeze/thaw cycles are not recommended, to mini-
commonly consist of the application of corrections after statistical mize stability issues. Instead, the samples should be initially ali-
analysis to address the multiple comparisons test problem (e.g., quoted in smaller volumes to preserve its quality for the analytical
FDR, Bonferroni), reduction of type II errors through power analysis purpose. At this point, the brand and even the lot/serial number of
calculations is a fairly solved problem in untargeted metabolomics. the sampling devices as tubes, vials should be carefully evaluated
This is mostly due to the unknown number of variables obtained in since it may be a source of contaminants such as plasticizers (e.g.,
a specific experiment, and the heterogeneity in the variance of the PEG or other polymers) that can cause serious signal interference.
different metabolites within and between each sample group.
Furthermore, power analysis calculations are dependent on the 2.3. Quality assurance
statistical test employed. As a good practice, we recommend a
sufficient sample size for adequate testing of normal distributions Quality assurance (QA) and quality control (QC) are essential
and performance of parametric tests, since rank-based tests per- processes defined as a set of mandatory procedures applied to
formed in low sample sizes vastly introduce type II errors. We also ensure the performance and quality in each step of the
Fig. 1. Experimental design of the GC-MS metabolomic experiment.
4
metabolomics workflow. Several recent reviews covered the best samples [24]. The ternary mixture isopropanol/acetonitrile/water
practices and proposed the standard operating procedures and (3:3:2 v/v/v) proposed by Fiehn's group [4] represents another
community consensus on improving and systematizing QA/QC adequate strategy. From our experience in GC-MS untargeted
practices for untargeted metabolomics studies [10e12,19]. The QA metabolomics, pure acetonitrile or pure methanol are very good
process includes a set of recommendations applied to the pre- alternatives for specific sample matrices. However, solvent extrac-
analytical, analytical, and post-analytical phases in the experi- tion composition should be optimized for each type of sample.
mental design. QA/QC procedures refer to the actions to be taken in Even though many metabolites are amphoteric in nature and
order to demonstrate the quality of the acquired data. Pre- acidic/basic at different pH, we do not recommend any pH
analytical considerations are essential to consider the standardi- adjustment for minimizing bias in untargeted metabolomics.
zation of adequate sample collection and handling protocols. Regarding chemical derivatization, the process involves two steps:
Regarding analytical QA, a “blank” sample preparation process (i) methoxymation for protection of the alpha-keto group and (ii)
should be always included by applying the same solvents, chem- silylation, usually with N,O-Bis(trimethylsilyl)trifluoroacetamide
icals, consumables, and standard operating procedure as for the (BSTFA) or N-Methyl-N-(trimethylsilyl)trifluoroacetamide (MSTFA)
test samples, but in the absence of any actual biological sample in for converting acid protons into their trimethylsilyl-derivatives.
GC-MS. That will allow us to identify artifacts and different sources Additional information is described elsewhere [7]. BSTFA/MSTFA are
of contamination. In addition, Quality control samples (QCs) should more reactive than N-tert-butyldimethylsilyl-N-methyltri-
be obtained, treated and analyzed with the study samples. QCs fluoroacetamide (MTBSTFA). It is demonstrated that increasing the
should be homogenous and representative of the qualitative and bulk of substituents on the silicon atom and the resulting steric hin-
quantitative composition of the study samples [20]. For such pur- drance, suppress derivative formation (MTBSTFA and hexoses) [25].
pose, the most commonly applied practice is to combine and mix The authors have developed matrix specific protocols to prepare
equal volumes from each of the experimental samples. Alterna- samples for metabolite analysis by GC-MS that emphasize the
tively, for studies with large number of samples, QCs can be pre- central importance of incorporating reagent blanks and QC samples
pared from a smaller but representative number of samples. within the workflow. This ensures continuous evaluation of
Already-prepared QCs should be aliquoted to minimize freeze- contamination and operational factors across all aspects of an
thaw cycles. Considering the highly volatile nature of already analytical campaign (see supplementary materials).
derivatized metabolites for GC-MS analysis, each QC sample should Addition of internal standard (IS) is always recommended to
be analysed only once. QCs should be included in the sequence minimize errors from the drift response and from the methodology.
ideally every six study samples. Repeated analysis of samples that Regarding the use of stable isotopes, metabolomics with ultra-
have already been injected into the GC-MS does not result in high-resolution MS based on stable isotope strategies are mainly
reproducible data, especially for volatile compounds [10e12,19] focused on metabolic flux for quantification, identification and
The calculation of RSD for each metabolic feature for all QCs across pathway analysis [26]. In untargeted metabolomics we encourage
the analytical batch provides a quantitative measurement of pre- the use of at least two IS, called IS1 and IS2. We recommend tri-
cision. It was suggested that data should be filtered out based on cosane (C23H48) as IS2, because its peak abundance can be used for
RSD higher than 30% in GC-MS data [10,12]. Moreover, molecular the determination of drifts in the system response and for the
features showing a comparable variation within the QC samples evaluation of the injection process performance. Tricosane is a
relative to the variability within the biological samples of interest in volatile compound not present in biological samples, soluble in
the study can also be considered unreliable and thus removed from heptane and unable to be derivatized by the above-described
subsequent statistical analysis. reagents.
Our protocols are continuously revised and updated. Then, after
3. Sample treatment before GC-MS analysis evaluating a set of possible candidates, we recommend as IS1 an
isotopically labelled IS, d31-palmitic acid (d31-C16:0). This com-
Sample treatment before their analysis by GC-MS encompasses: pound offered the best performance-price ratio when compared to
(i) homogenization and sample preprocessing, (ii) protein removal other candidates. Besides, palmitic acid does not give multiple
and extraction, and (iii) chemical derivatization. At the initial step, derivatives. Considering the parameters of the analytical method,
samples must be thawed on ice to minimize enzymatic reaction the retention time of the single derivative observed is 18.6 min.
rates before metabolic quenching. Extraction solvents must be kept Moreover, the repeatability of the sequence increases after
at low temperature (20 C) prior to their use for causing enzyme normalization by this IS. In addition, its price is lower than other
denaturation and subsequent cessation of metabolism. Even related substances (e.g. d27-myristic acid). Besides, 4-nitrobenzoic
though sample extraction with non-polar solvents has been acid is also a good choice for IS1 when analyzing urine or amniotic
described since the fifties and even recently for stepwise liquid- fluid.
liquid extraction [21] whenever possible, avoiding the use of non- The protocols are organized according to the following sample
polar solvents as tert-butyl methyl ether (MTBE) or chloroform is types:
recommended, because these solvents exhaustively extract non-
polar lipids such as triglycerides that are non-volatile and can - Biofluids: plasma, urine, breast milk, amniotic fluid, bron-
contaminate the inlet and guard column. In our experience, the choalveolar fluid.
best alternative approach is the use of mid-polarity solvents in a - Tissues: brain, lung, kidney, liver, optic nerve.
single-phase extraction to avoid the split of metabolites into two - Cells and mitochondria: adipocytes, macrophages, leishmania
immiscible liquid phases. Pure methanol or acetonitrile are known promastigotes, mitochondria.
to be less efficient for extraction of complex lipids. Comparing - Culture media.
methanol with acetonitrile, salts are more soluble in methanol and
that could lead to precipitation problems in the GC vial, depending For illustration and better understanding of the sample treat-
on the type of sample. Besides, methyl esters of fatty acids could be ment process, a tutorial video entitled: “GC-MS sample prepara-
detected as artifacts promoted by the slightly acidic nature of tion.m2v” is linked to the written manuscript.
methanol [22]. However, methanol was found one of the best
extraction solvents for metabolomics [23] and especially for urine
5
3.1. Detailed sample treatment protocols for biofluids Urine, Amniotic Fluid [27]
Plasma/Serum, Bronchoalveolar lavage fluid [27] Sample collection

Urine: Consider type of urine sample e.g., 24 h, first morning.
Sample collection Antibacterial preservative is not recommended. Instead, samples
Plasma: Collect whole blood from a peripheral vein directly into should be immediately stored at low temperature (80 C).
tubes containing anticoagulant (e.g., Na2EDTA). Mix the tube Amniotic Fluid: Sample collected during amniocentesis into
thoroughly after collection in the blood collection tube (for sterile container.
example vacutainer), prior to centrifugation ~ 4 C, 20 min, 2,000g. Samples should be centrifuged at 10,000g for 10 min at 4 C (to
The clear, upper layer is plasma. Citrate-anticoagulated plasma is remove particulates).
not recommended because endogenous citrate concentrations
could not be compared, so this crucial information would be lost. 1. Homogenization and sample preprocessing TIMING ~ 1 h
Serum: Collect blood samples from a peripheral vein into the
serum collection tube with clotting activator (silicate). Shake the Thaw samples on ice and vortex-mix for 1 min. High concen-
vial gently but thoroughly, and store at controlled temperature trations of urea can cause chromatographic interference and mask
(4 C) in an upright position to allow coagulation. Clot formation many low-intensity metabolites. Therefore, enzymatic reaction
should be completed after 20e30 min in most cases. Centrifuge to with urease is used to deplete urea. Add 30 units (50 mL) of urease to
separate the serum from the blood clot ~ 4 C, 10 min, 2,000g. 200 mL of urine and incubate at 37 C in a stove for 30 min to
Bronchoalveolar lavage fluid (BALF): collect the samples from decompose and remove excess urea. Urease solution should be
lungs into sterile tubes and centrifuge ~ 4 C, 10 min, in slow freshly prepared and mixed frequently to avoid variable addition
speed ~ 250e300g. due to enzyme sedimentation.
Amniotic Fluid: In early gestation, amniotic fluid is a dialysate
1. Homogenization and sample preprocessing TIMING ~ 1 h that is identical to the fetal and maternal plasma, but with a lower
protein concentration. By the second trimester, the fetus contrib-
Thaw sample on ice and vortex-mix the sample for 2 min. utes to amniotic fluid volume and composition almost exclusively
through urination.
2. Deproteinization and metabolite extraction TIMING ~ 1 h
On ice, pipette 40 mL of plasma to a plastic conical tube
(Eppendorf ® tube) and add 120 mL of pre-cooled acetonitrile On ice, add 800 mL of pre-cooled methanol (20 C) containing
(20 C) containing 25 ppm of d31-palmitic acid (IS1). Vortex-mix 25 ppm-4-nitrobenzoic acid as IS1. Vortex-mix for 5 min and
for 2 min and incubate the samples on ice for 20 min. Vortex-mix incubate the samples on ice for 20 min. Vortex-mix for a few sec-
for 1 min and centrifuge the sample at 15,400g for 10 min at 4 C. onds and centrifuge the samples at 15,400g for 10 min at 4 C.
3. Evaporation TIMING ~ 2 h 3. Evaporation TIMING ~ 2 h
Transfer 100 mL of the supernatant to a GC vial equipped with a Transfer 200 mL of the supernatant to a GC vial equipped with a
200 mL insert and dry the extracts in a speed vacuum concentrator 200 mL insert and dry the extracts in a speed vacuum concentrator
at 30 C to complete dryness. at 30 C until complete dryness.
The conditions for methoxymation, silylation and set up for The conditions for methoxymation, silylation and parameter set
analysis are detailed in Table 2. up for analysis are detailed in Table 2.
Table 2
Derivatization procedure for each type of sample.
SAMPLE TYPE SAMPLE METHOXYMATION SILYLATION SET UP FOR

ANALYSIS
TIMING ~0.5 h þ 16 h TIMING ~ 2 h TIMING ~

20 min
BIOFLUIDS Plasma/Serum/ Add of O-methoxyamine hydrochloride (15 mg/mL) in Add of BSTFA with 1% TMCS to each sample. Close the Centrifugation
Bronchoalveolar10 mL pyridine. Close the vials and vortex-mix 10 mL vials and vortex-mix for 5 min. Place the GC vials for 15 min at
lavage fluid vigorously for 5 min, ultrasonicate for 2 min and into an oven for 1 h at 70 C for silylation. Cool the 2,500g at 20 C
vortex for 2 min e repeat the process 3x. Cover samples for approx. 15 min at room room
with aluminium foil and incubate under darkness temperature in the dark.
Urine/Amniotic fluid Add at room temperature for 16 h. Add Add 100 mL of n-heptane containing 20 ppm of
30 mL 30 mL tricosane (IS2) to each GC vial. Vortex mix for
Breast milk Add Add 2 min.
10 mL 20 mL
TISSUES Brain/hippocampus, Add Add
lung, kidney tissue, 20 mL 20 mL
liver, pancreas, and
optic nerve
CELLS AND Adipocytes, Add Add
MITOCHONDRIA Macrophages, 10 mL 10 mL
Leishmania
promastigotes and
Mitochondria
CELL CULTURE Culture media Add Add
MEDIA 10 mL 10 mL
6
Breast milk [28] 10 min at 4 C. (#) MTBE was added for a multiplatform strategy,
but if only GC-MS is required, then skip this addition.
Sample collection
Sample is collected using manual breast pump and stored 3. Evaporation TIMING ~ 2 h
at 80 C.
Transfer 300 mL of the supernatant (80 mL for optic nerve) to a GC
1. Homogenization and sample preprocessing TIMING ~ 1 h vial equipped with a 300 mL insert. Dry the extracts in a speed
vacuum concentrator at 30 C until complete dryness.
Thaw samples on ice and vortex-mix for 2 min. The conditions for methoxymation, silylation and parameter set
up for analysis are detailed in Table 2.
Take 50 mL of milk sample, add 175 mL methanol with 25 ppm of 3.3. Detailed sample treatment protocols for eukaryotic cells and
d31-palmitic acid (IS1), add 175 mL MTBE. Vortex-mix 1 h. Centri- mitochondria
fuge the samples at 4,000g for 15 min at 15 C.
Cell pellets should have at least 106 cells (ideally 107). Remove
3. Evaporation TIMING ~ 2 h completely the culture medium and wash several times. Remove
liquid. Then, freeze the pellet in liquid nitrogen and store it
Transfer 150 mL of the supernatant to a GC vial equipped with a at 80 C.
200 mL insert and dry the extracts in a speed vacuum concentrator
at 30 C to complete dryness. Adipocytes [35]
The conditions for methoxymation, silylation and parameter set
1. Homogenization and sample preprocessing TIMING ~ 2 h
Add 350 mL pre-cooled methanol (-20 C) with 25 ppm of d31-

3.2. Detailed sample treatment protocols for tissues [29e33]
palmitic acid (IS1) on dry ice and vortex for 1 min. Three
freezeethaw cycles are needed for complete cell disruption, allow
Brain/hippocampus, lung, kidney tissue, liver, pancreas, and
to stand for 10 min at 4 C, place in liquid nitrogen for 10 min and
optic nerve
thaw in an ice bath for 10 min with brief vortex-mixing. For
eukaryotic cells (at least for monolayers) a sampling option is to
Sample preparation
add solvent (cold 80% methanol) directly to washed cells.
Representative tissue samples should be taken, and blood con-
tact with the organ surface must be minimized. For brain tissue
2. Deproteinization and metabolite extraction TIMING ~ 1.5 h
analysis, dissect the targeted area for the analysis. If it is not
possible to dissect a specific area, collect the entire right or left
Centrifuge at 5,725g for 5 min at 4 C, extract the supernatant
hemisphere.
and re-extract the pellet again with 350 mL of cold methanol with
Recommendations to avoid postmortem-induced changes in
IS1. Centrifuge at 5,725g for 5 min at 4 C and extract the super-
brain samples after tissue collection were previously investigated
natant. Mix both extracts and vortex-mix for 20 s.
[34]. Tissue samples must be snap-frozen in liquid nitrogen and
stored at 80 C until analysis.
3. Evaporation TIMING ~ 2 h
Evaporate 200 mL of the mixed supernatant to dryness at 30 C.
The conditions for methoxymation, silylation and parameterset
Immerse the samples in Liquid N2. Weight frozen tissue
(~30 mg), add cold methanol:H2O (1:1) v/v to the sample in a ratio
10 mg tissue:100 mL solvent (except for optic nerve). Consider using
Macrophages [36]
ultrasonic lysis. Intensity of sonification: time: 30 s; amplitude: 50;
cycle: 0.8; on ice and/or bead mill - QIAGEN Tissue Lyser (Qiagen,
Sample treatment
Germany). Place TissueLyser LT adapter on ice for at least 15 min.
Cell pellets containing approximately 107 cells.
Add 2.8/5 mm diameter stainless steel beads for soft tissue or 7 mm
for hard tissue to each tube. Perform 3 cycles for 5 min at 50 Hz on
ice. The number of the cycles, speed and time could vary depending
on the tissue type.
Thaw cell pellet. Perform double extraction by sequential
For optic nerve add 100 mL of cold MeOH (20 C) with 25 ppm
addition of (1) MeOH:MTBE(#) (4:1, v/v) with 25 ppm of d31-pal-
of d31-palmitic acid (IS1) and 2 mm particle size glass beads for the
mitic acid (IS1) and (2) MeOH:H2O (4:1, v/v) (500 mL each per
TissueLyserLT homogenizer (Qiagen, Germany), and place them on
fraction of 4 107 cells). Each solvent addition was followed by
ice between cycles for 30 s, 6 times each. Centrifuge for 10 min,
three cycles of freeze/thawing (liquid N2/cold-water, 10 s), soni-
13,000g at 4 C.
cation (15 W, 6 min), vortex-mixing (1 min), and centrifugation
(16,000g, 10 min, 4 C).
2. Deproteinization and metabolite extraction TIMING ~ 1 h (not
for optic nerve) (#) MTBE addition for multiplatform approach, otherwise skip
this addition.
On ice, pipette 100 mL of tissue homogenate (150 mL for liver and
pancreas) to an Eppendorf ® tube and add 320 mL of pre-cooled 2. Deproteinization and metabolite extraction TIMING ~ 0.5 h
methanol (20 C) containing 25 ppm d31-palmitic acid as IS1.
Vortex-mix for 20 min and centrifuge the samples at 4,000g for Combine equal volumes from fractions (1) and (2), then vortex
7
mix and aliquote for subsequent analysis. 3.4. Detailed sample treatment protocols for culture media [42]
3. Evaporation TIMING ~ 2 h 1. Homogenization and sample preprocessing TIMING ~ 0.5 h
Transfer 200 mL of the supernatant to a GC vial equipped with a Thaw on ice, vortex-mix 1 min. Add cold acetonitrile with d31-
200 mL insert and evaporate to dryness at 30 C. palmitic acid 25 ppm (IS1) to the culture medium in a ratio 3:1 (v/
The conditions for methoxymation, silylation and parameter set v). Vortex-mix for 2 min.
2. Deproteinization and metabolite extraction TIMING ~ 0.5 h
Leishmania promastigotes [37e40]
Centrifuge at 15,700g for 20 min at 4 C and collect the
Sample treatment supernatant.
Cell pellets containing approximately 4 107 promastigotes of L.
donovani, L. infantum, L. amazonensis or L. braziliensis. 3. Evaporation TIMING ~ 2 h
1. Homogenization and sample preprocessing TIMING ~ 2 h Transfer 100 mL of the supernatant to a GC vial equipped with a
200 mL insert and evaporate to dryness at 30 C.
Thaw the pellet. Add 350 mL methanol:water:chloroform 3:1:1 The conditions for methoxymation, silylation and parameter set
(v:v:v) at 4 C followed by disruption in a Tissuelyser LT (2 glass up for analysis are detailed in Table 2.
balls, 2 mm, 20 min, 50 Hz), for L. amazonensis and L. braziliensis,
using 50 mg of 426-600 mm diameter glass beads for 10 min.
4. Analytical method
2. Metabolite extraction TIMING ~ 20 min
In this section, the chromatographic method parameters will be
described as well as those of data acquisition for low- and high-
Centrifuge at 15,700g for 15 min at 4 C.
resolution mass detectors. Moreover, calibration of the MS system,
experimental set up, worklist design, retention time locking RTL
3. Evaporation TIMING ~ 2 h
tool will be detailed. Practical recommendations for the best
maintenance of the GC-MS instrument and considerations for long
Transfer 200 mL of the supernatant to a GC vial equipped with a
sample batches will also be covered.
A) Low-resolution equipment: GC-(EI)-Q-MS with an auto
sampler (Agilent Technologies 7693). Gas chromatograph
(Agilent Technologies 7890A GC with split/splitless injector)
Mitochondria [41]
coupled to a Mass Selective Detector (Agilent Technologies
5975 inert MSD with Quadrupole Detector).
Sample treatment
B) High-resolution equipment: GC-(EI)-QTOF-MS system with
After 36 h incubation, detach the cells from the culture plates
auto sampler (Agilent Technologies 7693), gas chromato-
and pellet by centrifugation at 1,000g for 5 min. Wash the cells once
graph (Agilent Technologies 7890B GC with split/splitless
with 25 mL of phosphate-buffered saline. Collect the cells by
injector) coupled to an Accurate-Mass Q-TOF detector (Agi-
centrifugation at 1,000g for 5 min, and then snap freeze and store
lent Technologies 7250).
the cell pellet at 80 C. Isolate mitochondria using a Mitochondria
Isolation Kit for Cultured Cells following the manufacturer's
Chromatographic method (GC-Q-MS and GC-QTOF-MS)
instructions.
Helium (99.9999%) carrier gas. Flow rate: 1 mL/min.

1. Homogenization and sample preprocessing ~ 0.5 h
Injector T: 250 C.
Split ratio 10:1 (12:1 for GC-QTOF-MS system). For urine 2:1
Thaw isolated mitochondria on ice and homogenize them by
(4:1 for GC-QTOF-MS system). If the number of cells is not >107,
vortex-mixing for 1 min. Subsequently, centrifuge the sample at
we recommend trying two injections of the same QC sample at
16,000g and 4 C for 15 min) and discard the supernatant.
different split ratio 10:1; 4:1; 2:1 (12:1; 6:1 for GC-QTOF-MS
systems) and select the best split ratio according to the ob-
2. Deproteinization and metabolite extraction ~ 2 h
tained profile. Compare the signals of lactic acid-2TMS (6.7 min),
urea-2TMS (9.3 min) and citric acid-4TMS (16.4 min).
Re-suspended the remaining pellet in 100 mL of water, vortex-
Column: DB5-MS (30 m length, 0.250 mm i.d., 0.25 mm film
mix for 1 min and lyse mitochondria by four cycles of
thickness, 95% dimethyl/5% diphenylpolysiloxane) with a pre-
freezeethaw (frozen in liquid nitrogen and thawed at 40 C). To this
column (10 m J&W integrated with Agilent 122-5532G)
solution, add 300 mL of pre-cooled acetonitrile (-20 C) with 25 ppm
Oven temperature program: 60 C initial for 1 min to 325 C at
of d31-palmitic acid (IS1). Vortex-mix the mixture for 1 min and
10 C/min, hold 10 min.
centrifuge (16,000g and 4 C for 10 min).
Equilibration time: 2 min.
Total time of analysis: 37.5 min.
3. Evaporation ~ 2 h
RT locked at 19.663 min (elution time of methyl stearate
C19H38O2, IS for system suitability).
Transfer 200 mL of the supernatant to a GC vial equipped with a
GC- Q-MS 5975 system. MS parameters
8
Detector transfer line, filament source and quadrupole tem- not recommended) and then (v) QC samples must be injected
perature set at 280 C, 250 C and 150 C, respectively. equally distributed along the sequence to monitor response drift or
Ion source (EI): 70 eV. any instrument error (one injection per vial). The sequence will
MS scan mode: 2 spectra/s. start with another QC, then, to correct analytical variability, sam-
Mass range: 50e600 m/z. ples in a randomized order based on an equal distribution of con-
MS Tuning: Perform autotune after periodic maintenance trols and case samples along the sequence. The QC samples should
following the instrument manual. Verify tune results. be injected, if possible, at least every six samples. The sequence will
Verify the sensitivity of the system by injection of the same finish with the last QC followed by the last blanks, as these ones will
internal standard solution (methyl stearate 10 mg/L in n-hep- break the equilibrium state of the system.
tane, IS for system suitability). TIC signal must be > 450 kcounts, Parameters to be checked after injection of the initial IS:
if not increase the detector GAIN. Retention time of methyl stearate value and repeatability:
19.66 min. Otherwise, relock the method. Relocking consists in
GC-QTOF-MS 7250 system. MS parameters injecting the solution of this internal standard at different flow rate
until the right value of this flow rate is found to obtain the desired
Detector transfer line, filament source and quadrupole tem- retention time for the internal standard.
perature set at 280 C, 250 C and 150 C, respectively. Peak abundance: TIC > 450 kcounts (injection volume: 2 mL (GC-
Ion source (EI): 70 eV. Q-MS), 1 mL (high resolution GC-MS), and good repeatability CV
MS scan mode: 10 spectra/s. <10%.
Mass range: 50e600 m/z. Tailing: Cut 20 cm of the guard column only if a symmetric peak
MS Tuning: Perform fast tune following the instrument manual. shape is not obtained.
Verify the intensity of the base calibrant peak.
Mass calibration: Perform mass calibration every 6 injections. 4.2. Sequence design
Verify the sensitivity of the system by injection of the same
internal standard solution (methyl stearate C19H38O2, 10 mg/L in Create a new sequence with the following injections (Table 3):
n-heptane, IS for system suitability). TIC signal must be > 450 Each day 40 injection can be performed according to the
kcounts. schedule sequence, so do not prepare more than 34 samples per
day.
4.1. Instrument set-up 4.3. Calibration of the instrument
Install a new liner and septum every new sequence and inject IS High-resolution GC-QTOF-MS
and QC several times to saturate the new liner and remove Every acquisition method includes a tunning file that must be
contaminant in the new septa, liner, and column. If the IS peak updated, performing a fast tune following the instrument manual
shape is not appropriated (e.g., observed peak tailing), cool the after source maintenance works or changing the column. When
system and cut 20 cm of the guard column. Do not forget to perform performing autotune of the MS system, tune parameters for both
annual system maintenance. standard energy (70 eV) and low-energy source (15 eV by default)
Every sequence should contain the first ~5 injections for system will be adjusted. Additionally, different values for Electron Energy
suitability. First (i), with methyl stearate solution in n-heptane and Emission Energy can be manually set in the source and save the
(10 ppm) to confirm the instrument's sensitivity, repeatability, and tune report.
efficiency. Then, injections will be continued (ii) with sample The equipment incorporates an ampoule containing a tuning
blanks to detect adsorption and contamination of the column, re- mix with 8 calibrants with m/z ranging from 69 to 614. The absolute
agents, or solvents, and (iii) reference mixtures: n-alkanes and and relative abundance of the calibrants, the accuracy in the
FAME (Fiehn's method). Subsequently, QC sample will be injected determination of their masses (always less than 2 ppm) simulta-
(iv) for equilibration of the system, 5 times for GC-Q-MS and 10 neously that the shape of the peak have to be verified in this tune
times for GC-QTOF-MS (>3 injections per vial for equilibration is file. During the analysis the mass accuracy in the detection is
Table 3
Sequence design before the analysis.
Sample type Characterization Nr of injections Time (approx.)
System suitability IS Methyl stearate (IS) 10 mg/L in n-heptane 5 200 min

Blank 1 and 2 Sample blank with IS1 and IS2 1 each 80 min.
Total time: 280 min (4.7 h)
Reference mix 1 Alkane mix as retention index markers across chromatograms for NIST library 1 40 min
Total time: 320 min (5.3 h)
Reference mix 2 FAME mix as retention index markers across chromatograms for Fiehn's library 1 40 min
Total time:360 min(6h)
QC 1 Quality control for equilibration of the system 5 (10 high resolution) 200/400 min
Total time: 560/760 min (9.3/12.7 h)
Study samples Control/Case group 1 each x 6 240 min
6 samples in random order Total time: 800/1000 min (13.3/16.7h)
QC 2 Quality control sample 1 40 min
Total time: 840/1040 min (14 h/17.3h)
Repeat the last two rows until the whole set of samples are analysed
Blank 3 and 4 Sample blank with IS1 and IS2 1 each 80 min.
9
guaranteed by including mass calibration in the sequence, this reactions, working as a tester of the analytical variability of the
calibration lasts a couple of minutes and is performed every 6 derivatization process.
injections. Visual inspection of the chromatograms allows the detection of
In the MS parameters of the analysis method, Standard or Low discrepant profiles due to different factors such as acquisition
Energy electron ionization (EI) mode may be selected for the problems, failed injection, and incorrect derivatization, among
analysis in full scan mode or for MS/MS mode as illustrated in Fig. 2. others. While the measurements are being performed, it must be
verified that the signal of the two internal standards, IS1 and IS2,
4.4. Checklist remains constant in all the samples. IS2 is used to monitor sample
injection and correct the intra-batch fluctuation of the RT of the
Before running any study samples, check several aspects to in- metabolites, it is not derivatized so the allowance in the threshold
crease the reliability of the analysis. Fig. 3 depicts all these points of RSD value is severe, usually 5% in the same matrix, sample and
regarding samples, system, sequence and analytical performance. QCs. However, IS1 works as a tester of the analytical variability of
the derivatization process, and its RSD is approximately 10%. We
4.5. Troubleshooting must also confirm that the QCs samples that are being analyzed
periodically tightly overlap. This double-check highlights the inter-
Table 4 includes several troubles that can occur when analysing analytical quality of our experiment.
samples by GC-MS, regarding baseline, peaks, and retention time,
potential trouble source and solutions. 5.2. Preprocessing data: deconvolution and searching against
spectral libraries
5. Data extraction
Once the sequence has been completed, we would proceed to
5.1. Quality assurance and samples for quality control the preprocessing stage, starting by the deconvolution and then
searching against spectral libraries.
When quality control (QC) samples have been prepared as a pool In metabolomics, deconvolution is the process of separating
of the same set of samples, the relative standard deviation (RSD) of computationally signals attributable to individual compounds that
each metabolite across the QCs is the best indicator of the repro- overlap in the retention time dimension. The results of the
ducibility and the repeatability of the analysis. deconvolution process are pure spectra associated to each co-
In addition, QCs clustering in multivariate analysis methods eluting compound, even for signals of less abundant compounds
such as principal components analysis, PCA, is a signal of low masked in the total ion chromatogram (TIC). In other words, after
analytical variability. the deconvolution, the ions corresponding to the same chemical
The two internal standards (IS1 and IS2), above mentioned, play species are grouped, enabling spectral similarity search for
an important role in any untargeted metabolomics study. As IS2 deconvoluted signals associated to potential compounds in spectral
does not need to be derivatized therefore is added to each sample at libraries.
the end of the sample treatment to monitor sample injection and to AMDIS is the best-known software to perform the deconvolu-
correct the intra-batch fluctuation of RT of the metabolites. How- tion process (automated mass spectral deconvolution and identi-
ever, a much better approach is to use IS1 that follows all the fication system) from NIST developed based on real EIC peaks [43].
Fig. 2. Parameters adjusted in the high-resolution GC-MS system for Standard or Low Energy EI mode.
10
Fig. 3. GC-MS analysis checklist for low- and high-resolution system.
Its great success probably relies in its free access but also tends to mzmine.github.io/ADAP_user_manual.pdf; https://cran.r-project.
produce many false-positive components [44], for this reason there org/web/packages/erah/vignettes/eRahManual.pdf].
are new approaches that combine AMDIS with another algorithms This tutorial focuses in a specific workflow optimized for
as, for example, Ratio Analysis of Mass Spectrometry (RAMSY) downstream data processing using Agilent MassHunter software.
which facilitates compound identification via comparison between We have chosen an AMDIS-based software commercially available
MS peak-intensities that form non-resolved chromatographic from Agilent, MassHunter, because it allows the dual approach to
peaks [45]. MS-DIAL [46] and ADAP-GC3.2 [47] are another tools find compounds by chromatogram deconvolution in an equipment
that perform deconvolution based on the grouping of the EIC peaks with low- and with high- resolution mass spectrometry analyzers.
of the fragments. However, additional series of software based on The software allows compound identification by MS library search
multivariate curve resolution (MCR) methods are appearing in a relatively easy, fast and accurate way without the need to be a
recently. eRah [48] and ADAP-GC 4.0 [49] belong to these last user with advanced computer skills.
groups that eliminate the step of selecting model peaks usually With Agilent software, deconvolution and spectral library
leading a better separation of coeluting components and missing search are performed with the same software namely, MassHunter
less low-intensity peaks but, on the other hand, taking significantly Quantitative Analysis Unknowns Analysis. Fig. 4 illustrates the
more time and producing model peaks that are usually very whole process with low and high resolution GC-MS systems:
different from real EIC peaks in shape. The file in Unknowns Analysis is called analysis and, as the first
Another approach to GC/MS data deconvolution is based on a approach, the samples including QCs should be added to be
model called PARAFAC2 (PARAllel FACtor analysis 2) [44]. This deconvoluted. The next step is the choice of the deconvolution al-
model allows the mass spectrum to be extracted for each compo- gorithm, as follows:
nent by combining information from multiple chromatograms
obtained for the same sample. This method requires that “at least (i) Deconvolution: If data results from a nominal mass device.
five samples with independent variations must be included in the (ii) SureMass: if the data is obtained from an accurate mass
sample set” and PARAFAC2 usage is limited to the mathematical instrument. This is a much more powerful and specific al-
users and needs extensive coding (usually in MATLAB). Individual gorithm for exact mass systems. In this last case, data must
tutorials and setting up of the data analysis pipeline for these tools have been acquired in profile format and subsequently
are readily available for non-familiarized users [https://mtbinfo- converted to SureMass format. This conversion can be per-
team.github.io/mtbinfo.github.io/MS-DIAL/tutorial.html; https:// formed while the data is recorded if specified in the
11
Table 4
Problems that can occur in GC-MS analysis affecting to the baseline, peak shape or retention time. Possible origins and solutions.
Practical troubleshootings - The most common problems
Problem Possible causes Solution
Identify problem Track events Revise recent changes in the system (instrument notebook and Logbook): column, septum, liner, syringe,
type of the samples, number of injections, method, maintenance e.g., column cut, detector cleaning, liner
changes, etc.
Basic check Check sample preparation, analytical condition, gas purity and gas flow, syringe, connections, inject test
mix (e.g., IS) and compare to previous data to ensure restored analytical performance
BASELINE
High column bleed Contamination (column/injector) e Trim or bake out the column
e Clean injector and detector e Replace the inlet liner e Solvent rinse the column
e Replace the column
Improper column conditioning e Increase conditioning time and/or temperature
Leak in system causing column e Check for leaks in the system
oxidation e Replace column
Unstable baseline Contaminated carrier gas e Replace carrier gas and/or detector gas filters
(Spiking, Noise, Drift) Contaminated injector or detector e Clean system, perform regular maintenance
Contaminated column e Bake out the column
e Solvent rinse the column
e Replace the column
Septum bleed e Replace septum
Decomposition of stationary phase e Check for leaks in the system
e Check matrix for compatibility with the column
Electronic malfunction e Revise the system
PEAKS
No Peaks Injection problems, defective or e Verify the sample volume in the syringe
plugged syringe e Clean or replace syringe
Detector problems e Check if detector is turned on
e Ensure detector is working properly
Column, instalation problem or e Re-install or replace column
broken column
Tailing peaks Contaminated inlet liner or column e Clean or replace inlet liner
e Bake out or replace the column
Installation issues e Verify that the column is cut properly
e Verify correct installation distances
e Minimize dead volume
Additional peaks Carryover of sample or contaminants e Verify the analysis time
from previous runs
Contaminants in current sample or e Inject solvent
solvent e Use high-quality solvents
Solvent Peak Broad Bad column installation e Re-install the column
Injector leak e Find and fix leak
Split ratio is too low e Increase split ratio
Carryover/Ghost peaks Contaminated syringe or rinse e Clean or replace syringe
solvent e Replace rinse solvent
Sensitivity loss Contamination of column and/or e Clean or replace inlet liner
liner e Bake out, solvent rinse or replace the column
Issues with the sample, e Check sample preparation
decpmposition, evaporation e Check if the sample vial is well closed
Injector leaks e Find and solve any leaks
Resolution loss Damage to column stationary phase, e Replace the column
excessive column bleed
Injector problems e Check for leaks, split ratio, purge time, liner
Irreproducibility Inconsistent injection e Check injection parameters, standardize
Baseline distrurbances e Check possible reasons and correct problems
Defective or plugged syringe e Clean or replace the syringe
RETENTION TIME
Shift Change of solvent e Always use the same solvent quality for preparation of samples and standards
Leaks e Check for possible leaks
e Check injector
e Replace critical seals (e.g., septa, O-rings, etc.)
Analyte adsorption e Maintain inlet liner and GC column
Significant loss of stationary phase e Ensure run conditions
due to column bleed e If necessary, replace column
Decrease in gas flow rate e Inject an internal standard to determine the linear gas velocity. Adjust gas pressure to obtain proper
values for the analytical method
Column (contamination or loss/ e Clean or replace inlet liner
damage of stationary phase) e Bake out, solvent rinse or replace the column
12
Fig. 4. Data inspection, deconvolution, library search, data integration and data matrix for high resolution and low resolution GC-MS systems (up and down rows, respectively).
corresponding acquisition software or later with the Mass- library search in the following order: in-house library, then Fiehn's
Hunter Unknowns Analysis or MassHunter Quantitative library, and lastly NIST library.
Analysis software. Be aware of the disadvantage that the file Ideally, in the case of data generated from an accurate mass
sizes in profile format is large, and needs almost double instrument, the search should be performed in an exact mass li-
space compared to a standard data file. brary. Even better it is to build a personal compound database and
library, PCDL by injecting standards into our equipment with the
The mass spectra can be recorded in profile mode (also called same method. In this way, the exact mass spectrum and retention
continuum) but are often ‘centroided’. Centroiding is a process of time could be acquired. Unknowns Analysis allows for searching in
peak detection for a profile mode mass spectrum (hence in the m/z libraries in the following formats:.L, mslibrary.xml, and.cdb (PCDL
dimension, not in the chromatographic dimension)da gaussian format). Besides, the PCDL can also be used even if the experi-
region of a continuum spectrum with a sufficiently high signal-to- mental data has nominal mass.
noise ratio is integrated to give a centroided mass (a “stick” in the In addition, for each search, it can be specified if the library
mass spectrum as opposed to a continuous signal) and integrated contains retention time and if so, it can be selected whether the
area under the curve. This results in data of reduced size [50]. If the retention time penalty function applied is going to be considered
data was acquired only in centroid format, subsequent conversion Gaussian or trapezoidal. We recommend using the trapezoidal
to SureMass is not possible. With this format, the only algorithm function and 6 s for the retention time range and 5 s for the penalty-
available is deconvolution, increasing the time of analysis to such free RT range. The RT mismatch penalty can be considered multi-
an extension that almost prevents its application. plicative or additive, suggesting as a good choice multiplicative and
The parameters that define the deconvolution, both the stan- 45 s.
dard procedure with the Agile algorithm and the SureMass al- If the library did not have retention time but retention index, the
gorithm are identical and will be set in the Unknowns Analysis. software allows us to include a calibration file in which we collect
Firstly, m/z e.g., 28, 43, 44 coming from e.g., N2, CO2 must be the retention index values and the corresponding experimental
excluded in the deconvolution process. Secondly, it becomes retention times for a series of standard reference compounds that
fundamental to fix when to skip extract ion chromatograms, EIC, cover the whole range of experimental times (e.g., n-alkane for
peaks with signal-to-noise ratio below a threshold, selecting 1 as a specific for NIST library and a mixture of Fatty Acid Methyl Esters e
first approach to avoid any signal loss. Fig. 5 illustrates the whole FAME for Fiehn's library) since Fiehn's library includes the retention
process involving deconvolution and searching against libraries index on the FAMES scale and NIST that based on the n-alkane scale.
with SureMass algorithm. Thus, it is a highly recommended practice, especially while working
The retention time window size factor is the key parameter that on unknown compound identification, to include at the beginning
controls the grouping of EIC peaks into components. Low values are of the sequence the injection of commercial mixtures of FAME and
adequate for low abundant compounds and more components will n-alkanes allow us to build calibration files that will allow con-
be distinguished in the same range of time. Unknowns Software verting retention time into retention index through polynomial
allows the introduction of different values for this parameter, a good regressions.
option is 25, 50, 75, 100, and 200. A threshold in the sharpness of the The minimum m/z value can be set to compare the spectra,
peaks can also be included, computing component shape by aver- eliminating very low m/z that usually give misidentification (an
aging the shapes at the top of the sharpest peaks, i.e. 35 %. adequate selection is considering m/z > 50). Additionally, a mini-
The next tab allows for fixation of the parameters for the library mum match with the library spectrum can be set (e.g. 40%). Unlike
search, allowing for seeking in multiple libraries. If the "stop when the retention time which can be set independently for each library,
found" option is selected, the search order is very important, having the minimum m/z value and minimum match threshold must be
to place first the libraries that contain the compounds that are most fixed commonly to all the libraries.
likely to appear in our samples. We recommend performing a The whole process is fast for data including nominal mass, even
13
Fig. 5. Screen windows for deconvolution and compounds searching against libraries with the high-resolution system.
if large sample batches are included, and while searching in several 6. Relative quantification of components and unknown
compound libraries. However, computation times are dramatically compounds with low and high resolution MS detectors
increased when deconvoluting data with accurate mass. For such
kind of data, we recommend that Unknowns Analysis is executed The last step of the process is to Quantify these compounds in
only for QCs or, additionally, some representative samples from the whole set of samples, including blanks and QCs. This stage is
each group of samples. performed with Quantitative Analysis software. The option is called
At the end of the data reprocessing with the Unknowns Anal- batch and, here, all the experimental samples, blanks, and QCs have
ysis tool, a data table containing the sample names, the number of to be considered.
features, and the number of hits (compounds annotated in the li- A method for processing can be built in the Quantitative Anal-
brary search) will be presented (Fig. 6). ysis software based on this last CEF file, but the ions (qualifiers and
The compounds that appear are not necessarily the same in all target ion) that are selected based on experimental data have to be
samples but, generally, the interest is in quantifying all the com- carefully revised manually. This step is crucial for further process-
pounds that have been identified. For this purpose, a script in Un- ing. The ideal procedure is constructing a sub-PCDL from a selected
knowns Analysis can be run that enable to export all the hits with PCDL including all the compounds of this CEF file. This sub-PCDL
the highest match, creating a CEF file for each sample. Next, with can be built with the PCDL manager or using the Library Editor
the help of the Mass Profiler Professional software, it is possible to after converting this PCDL into an exact mass library. Then, when
align all the CEF files and generate a unique CEF file that collects all the method of the Quantitative Analysis is created based on this
the different hits of all the samples. This CEF file gathers all the PCDL, consequently, the target and qualifier ions certainly belong to
compounds that are going to be quantified; therefore, the align- the compound and will also have the corresponding exact mass. For
ment is performed with no restrictions. data with nominal mass, this process can be done from any library.
14
Fig. 6. Screen windows with the components and hits found in a sample.
It is very important to consider that when the Quantitative lack of exact mass libraries. In those cases, it would be necessary to
Analysis method based on the sub-PCDL or the sub-library is confirm the target ion with accurate mass with manual de-
created, the option weighed is chosen for selecting the target ions terminations that greatly lengthen the process, for example with
and not the most abundant ones, to avoid picking m/z 73 and 147. NIST and the tool named MS Interpreter.
The ion 73 is a fragment that appears in any compound that is After all the manual revision of the signal integrations an Excel
silylated during the derivatization process as correspond to the file can be exported from MassHunter Quantitative Analysis with
cation (CH3)3Siþ, being frequently the base peak. The ion 147, the areas of all the compounds and retention times in the whole set
(CH3)2¼Oþ-Si(CH3)3 appears whenever the compound has at least of samples. The first process to be performed after exporting the
two acidic hydrogens, the situation also very common that often data matrix is blank subtraction to remove the signals coming from
makes 147 the base peak. For these reasons, these two ions are non- the reagents themselves or the derivatization process.
discriminating ions and cannot be used as target ions in silylated The different chemical nature of the acidic proton replaced in
compounds. Consequently, it must be checked in all the com- the silylation reaction can be incomplete leading to multiple de-
pounds that the target ion is the right one for quantification. When rivatives due to single, double, triple, or even greater replacing.
the method is created, the qualifiers that are extracted from the Despite the silylation reagent is added in excess the same com-
PCDL/library are also selected. One of these qualifiers can be used pound results in more than one derivative. Therefore, we propose
instead of the target ion if the automatic selection of the software is to sum the abundances of the all the derivatives from the same
not appropriate for the integration of this compound area. metabolite.
The mass window must be set depending on whether the data
comes from the QTOF or the single Q. Logically, this window should
6.1. Data normalization
be much smaller in the first case than in the second (20 ppm and
0.5 Da respectively). The appropriate retention time window
Normalization is especially useful in different scenarios. For
should be fixed according to the shift of our experimental values to
example, urine samples can be normalized by osmolality or creat-
the ones of the PCDL or library.
inine content, cell samples by the number of cells, tissue samples by
Even checking all these parameters thoroughly, it would be
their weight or the protein content, etc.
necessary to review compound by compound manually that the
Normalization is also recommended in the case of a long
automatic integration is correctly performed and, if not, correct the
sequence to reduce the intra- and inter-batch variability. Generally,
integration manually. Besides, possible sample overloaded or
normalization in GC-MS analysis is carried out with respect to the
saturation of the signal should be manually inspected. It the peak
IS.
shape is flat due to saturation, a different m/z must be selected.
Normalization based on IS2, in our case tricosane, would reduce
With high resolution instrument, isotopic peaks (e.g. Mþ1þ) could
the variation that depends on the analysis, but a better method uses
be frequently selected avoiding saturation. By far, the longest pro-
IS1, the internal standard that undergoes derivatization reactions,
cess would be to include all those compounds that have been found
in order to take into account the variability introduced by these
in an analysis with accurate mass but are absent in a PCDL, in an
reactions themselves (in our protocols 4-nitrobenzoic acid or d31-
exact mass library. Unfortunately, this is quite frequent due to the
palmitic acid, as detailed above). However, when internal standard
15
normalization does not satisfy the desired results, other normali- variable. A range of scaling methods has been applied in metab-
zation strategies can be applied, which rely on data interpolation olomics including auto-scaling, Pareto scaling, range scaling and
upon modelling of a fitting curve. These include LOESS regression VAST scaling [54].
[51], spline-fitting-based methods [52], or support vector regres-
sion [53]. 7. Metabolite Identification
6.2. Data filtering 7.1. Curation of tables
In metabolomics studies, two main strategies for data filtering As general rule, manual curation of tables with listed metabo-
are used: (i) filtering by presence, based on the percentage of lites should be done before the biochemical interpretation. General
presence of that metabolite in the samples of each group, with this rules to be followed:
protocol of working with GC-MS data this method this methodol-
ogy does not apply and (ii) filtering based on quality assurance - Multiple derivatives should be substituted by the name of the
procedures and the threshold of RSD value calculated for each metabolite and sum their abundances.
metabolic feature detected in QC samples, being the commonly - Leave just one row per metabolite.
accepted maximum value for GC-MS data of 30%. - Substitute the name of the derivative by its traditional name.
Finally, the last steps in the data processing procedure are data - Reagent by-products and excess of reagent must be deleted.
transformation and scaling, especially before multivariate statisti- - Delete all those metabolites with the same abundance than that
cal analysis. observed in blank samples.
- Unsound compounds for the biological sample being analyzed
6.3. Transformation or physiological conditions should be confirmed.
Transformation can be applied in metabolite abundances from

the data matrix so that the transformed abundances follow a 7.2. Potential and experimentally confirmed capabilities of the
normal distribution. A commonly applied method in metabolomics accurate mass detectors
is the performance of a log transformation before statistical anal-
ysis, which reduces right-skewed variable distributions. When one of the aims of the study is the characterization of the
metabolic profile, high resolution GC-QTOF-MS offers advanta-
6.4. Scaling geous features: higher selectivity from the high mass accuracy
(nominally <5 ppm error) and average 10-fold sensitivity as well as
In metabolomics, scaling is required to reduce the unbalanced enhance structural elucidation capabilities for unveiling novel
effect that high abundant metabolites have in multivariate models. compounds derived from the accurate mass and the use of PCDL
The absolute value of the abundance differences between samples with exact mass, although only trained personnel will be able to
of these metabolites is very high relative to low abundant metab- assure the result. The main disadvantage is the need for additional
olites. Thus, without scaling, the importance of low abundant me- hard drive memory and external servers enable to save and storage
tabolites cannot be appropriately assessed in conventional heavy files. Besides, the NIST MS Search tool can help in the
multivariate statistical modelling. Scaling methods divide each structural elucidation of novel metabolites, by offering the exact
variable by a factor, the scaling factor, which is different for each mass of m/z fragment of possible candidates, therefore the number
Fig. 7. Comparison of chromatograms and spectra with the Low Energy source. A1: Total Ion Chromatogram (TIC) obtained for a QC sample at 10 eV and 250 C in the source. A2:
spectrum for the peak eluting at 11. min corresponding to threonine 3TMS. B1: TIC obtained for the same QC sample at 30 eV and 250 C in the source. C: Extracted ion Chro-
matogram (EIC) at different T in the source at constant low Energy 30 EV.
16
Fig. 8. Representative total ion chromatogram (TIC) of a QC plasma sample analysed by a high-resolution GC-MS system. A) Total Ion Chromatogram. B) Expansion of the profile
from 5.5 to 15 min and C) Expansion from 15 to 29 min *Artifact.
of possible compounds will be much lower. lower energy values can be set, and obtained spectra could be
Alternatively, the 7250 GC-QTOF-MS system also present a Low compared. While comparing chromatograms obtained e.g., at 70,
Energy source. According to the fundamentals of mass spectrom- 30, 15, 10 eV at constant source temperature, the whole profile
etry, 70 eV is the set energy of the EI source because the frag- signal diminished markedly and the fragmentation of the analyte is
mentation is high and reproducible. With a Low Energy source, less noticeable. While the molecular ion remains detectable, non-
17
Table 5
List of metabolites found in the human QC sample by high-resolution GC-MS. TMS: Trimethylsilyl. MO: Methyloxyme.
number Name RT (min) number Name RT (min)
1 Boric acid, 3TMS 5.74 79 Pyrophosphate, 4TMS 14.73

2 3-Heptanol, TMS 6.11 80 5-Aminovaleric acid, 4TMS 14.95
3 2-Heptanol, TMS 6.21 81 Taurine, TMS 15.00
4 Ethanolamine, 2TMS 6.40 82 Xylitol, 5TMS 15.05
5 2-Hydroxypyridine, TMS (artifact) 6.52 83 Putrescine, 4TMS 15.69
6 Pyruvic Acid, N-MO, TMS 6.58 84 Orotic acid, TMS 15.70
7 N,N-dimethylglycine, TMS 6.59 85 Aconitic acid, 3TMS 15.71
8 Lactic acid, 2TMS 6.73 86 alpha-glycerophosphoric acid, 4TMS 15.85
9 Carbamic acid, N-TMS 6.83 87 Citric acid, 3TMS 16.05
10 Glycolic acid, 2TMS 6.96 88 Histidine, 2TMS 16.28
11 Valeramide, TMS 6.99 89 3-methyl-L-histidine, TMS 16.31
12 Valine, TMS 7.26 90 Hipoxantine, TMS 16.43
13 Alanine, 2TMS 7.42 91 Ornithine, 4TMS 16.48
14 Acetoacetic acid, TMS 7.48 92 Citric acid, 4TMS 16.49
15 Glycine, TMS 7.64 93 Citrulline, 2TMS 16.57
16 2-Hydroxybutyric acid, 2TMS 7.70 94 Ketohexose, N-MO, 5TMS 16.74
17 Oxalic acid, 2TMS 7.85 95 Dehidroascorbic acid, TMS 16.76
18 p-Cresol, TMS 8.15 96 Ketohexose, N-MO, 5TMS 16.79
19 Leucine, TMS 8.19 97 Hippuric acid, TMS 16.80
20 3-Hydroxybutyric acid, 2TMS 8.20 98 Myristic acid, TMS 16.84
21 2-Hydroxyisocaproic acid, 2TMS 8.21 99 1,5-Anhidroglucitol, 4TMS 16.84
22 Isoleucine, TMS 8.45 100 Ketohexose, N-MO, 5TMS 16.97
23 Proline, TMS 8.57 101 Ketohexose, N-MO, 5TMS 17.01
24 3-Aminobutyric acid, 2TMS 8.94 102 Adenine, 2TMS 17.05
25 Valine, 2TMS 8.95 103 Dehidroascorbic acid, 2TMS 17.07
26 Methylmalonic acid, TMS 8.96 104 Ketohexose, N-MO, 5TMS 17.11
27 2-Ketoisocaproic acid N-MO, TMS 9.00 105 Aldohexose, N-MO, 5TMS 17.16
28 Urea, 2TMS 9.33 106 Histidine, 3TMS 17.18
29 Benzoic Acid, TMS 9.50 107 Tyrosine, 2TMS 17.22
30 Serine, 2TMS 9.61 108 Aldohexose, N-MO, 5TMS 17.28
31 Caprylic acid, TMS 9.67 109 Glucuronic acid, 3TMS 17.30
32 Ethanolamine, 3TMS 9.71 110 4-Hydroxyphenyllactic acid, 3TMS 17.33
33 Leucine, 2TMS 9.80 111 Aldohexose, N-MO, 5TMS 17.34
34 Pipecolic acid, TMS 9.84 112 Dehidroascorbic acid, 3TMS 17.39
35 ortho-Phosphoric acid, 3TMS 9.89 113 Glucuronic acid, 4TMS 17.40
36 Glycerol, 3TMS 9.89 114 Aldohexose, N-MO, 5TMS 17.49
37 Acetylserine, TMS 9.97 115 Mannitol, 6TMS 17.63
38 Isoleucine, 2TMS 10.02 116 Tyrosine, 2TMS 17.63
39 4-Aminobutyric acid (GABA), 2TMS 10.10 117 Ascorbic acid, 3TMS 17.65
40 Proline, 2TMS 10.16 118 3,4-Dihydroxymandelic acid, 4TMS 17.64
41 Succinic acid, 2TMS 10.20 119 D-Glucitol (sorbitol), 6TMS 17.73
42 Glycine, 3TMS 10.28 120 Tyrosine, 3TMS 17.74
43 Glyceric acid, 3TMS 10.65 121 Ketohexose, 5TMS 17.89
44 Uracil, 2TMS 10.76 122 Aldohexose, 5TMS 18.08
45 Methylmaleicacid, 2TMS 10.82 123 1-Hexadecanol, TMS 18.12
46 Nonanoic acid, TMS 10.83 124 Pantothenic acid, 2TMS 18.27
47 Fumaric acid, 2TMS 10.86 125 N-acetyl-ornithine, 2TMS 18.35
48 Homoserine, 2TMS 11.02 126 D-Gluconicacid, 6TMS 18.37
49 Serine, 3TMS 11.05 127 Palmitoleic acid, TMS 18.52
50 Pipecolicacid, 2TMS 11.07 128 d31-palmitic acid, TMS (IS1) 18.58
51 Citraconic acid, 2TMS 11.16 129 Palmitic acid, TMS 18.63
52 Threonine, 3TMS 11.35 130 Dopamine, 4TMS 19.13
53 Methionine, TMS 11.81 131 myo-inositol, 6TMS 19.26
54 Methyl aminoacetate, 2TMS 11.82 132 Uric acid, 4TMS 19.28
55 Aspartic acid, 2TMS 12.00 133 Methyl Stearate (IS) 19.61
56 Homoserine, 3TMS 12.21 134 Heptadecanoic acid, TMS 19.61
57 Capric acid, TMS 12.37 135 3-indolelactic acid, 3TMS 20.08
58 Aminomalonic acid, 3TMS 12.44 136 Kynurenine, 3TMS 20.19
59 Malic acid, 4TMS 12.80 137 Tryptophan, 2TMS 20.24
60 Tetrol, 4TMS 12.80 138 Tryptamine, 2TMS 20.33
61 Aspartic acid, 3TMS 13.12 139 Linoleic acid, TMS 20.37
62 Methionine, 2TMS 13.13 140 alpha.-Linolenic acid, TMS 20.40
63 Pyroglutamic acid, 2TMS 13.15 141 Tryptophan, 3TMS 20.43
64 trans-4-hydroxy-proline, 2TMS 13.20 142 Oleic acid, (E)-, TMS 20.38
65 4-Aminobutyric acid (GABA), 3TMS 13.23 143 Oleic acid, (Z)-, TMS 20.46
66 Glutamic acid, 2TMS 13.26 144 Tryptamine, 3TMS 20.46
67 Phenylalanine, 2TMS 13.55 145 Stearic acid, TMS 20.61
68 Cysteine, 3TMS 13.55 146 Xanthurenicacid, 3TMS 20.77
69 Creatinine, 3TMS 13.57 147 L-Cystine, 4TMS 21.12
70 a-ketoglutaric acid N-MO, TMS 13.80 148 Tricosane (IS2) 21.16
71 3-Phenyllacticacid, 2TMS 13.84 149 D-Glucose-6-phosphate, N-MO, 5TMS 21.28
72 Ornithine, 3TMS 14.20 150 Arachidic acid, TMS 22.36
73 Glutamic acid, 3TMS 14.32 151 1-Monopalmitin, 2TMS 23.42
74 Phenylalanine, 2TMS 14.35 152 Inosine,4TMS 23.44
18
Table 5 (continued )
number Name RT (min) number Name RT (min)
75 Triethanolamine, 3TMS 14.37 153 Glycerol monostearate, 2TMS 24.87

76 5-Aminovaleric acid, 3TMS 14.42 154 Hexosylhexose 25.40
77 2-Aminoadipic acid, 2TMS 14.45 155 alpha-Tocopherol, TMS 27.10
78 Lauric acid, TMS 14.51 156 Cholesterol, TMS 27.49
241. After filtering by presence 119 and 179 compounds passed the
filter. Considering just compounds with low variability, CV <30%,
153 metabolites were considered with the high resolution system
instead of 71. When CV <20% was considered, 123 compounds
passed the criteria, more than double that of GC-Q-MS.
The higher capabilities of the high resolution GC-QTOF-MS
system in terms of sensitivity and repeatability have allowed the
comparison of at least 2-fold more metabolites than that of GC-Q-
MS.
8. Conclusions
The concept “one size fits all” is not always accurate in untar-
geted metabolomics due to the diversity of matrices, polarities, and
chemical properties of metabolites. That is the reason why pro-
tocols should be adapted to different types of samples. However, it
should be kept in mind that as long as conditions are reproducible,
the comparison among groups, the final aim, is possible. GC-MS
holds powerful capabilities for the small polar metabolites anal-
ysis (short-chain organic acids, amino acids, monosaccharides,
Fig. 9. Comparison of identification capabilities for low and high resolution GC-MS
instruments.
among others) that usually elute in the dead volume in reversed-
phase LC-MS. Moreover, due to the retention time reproducibility
and fragmentation patterns, the “known unknowns” identification
related fragments disappear and the associated spectrum becomes is straightforward based on libraries. Recently, the incorporation of
quite simple. Fig. 7 depicts this effect, with Total Ion Chromato- TOF mass spectrometers has provided better coverage and sensi-
grams obtained at 10 eV and 30 eV after injection of the same QC tivity adding new capabilities. Indeed, it helps better coverage and
sample can be compared (A1 and B1 respectively) as well as their sensitivity for the “unknown unknowns” identification based on
corresponding spectra for the peak eluting at 11.3 min (A2 and B2 the exact mass of the fragments. Finally, manual curation of both
respectively) that corresponds to threonine with 3-trimethyl metabolite annotations and relative quantification is essential to
groups (3TMS). Confirmation of the accurate molecular ion and provide accurate biological interpretations.
its fragments could result very helpful for the identification or
confirmation of metabolites in the profile. Besides, Extracted Ion Author contributions
Chromatogram, EIC, (m/z 320.1537) at different temperatures show
the dependence of the signal of the molecular ion with that F.ReS, D.D. and D.R. organized and managed the protocols,
parameter. Experiments with this tool when combined with lower F.ReS, C.G-R. M.F-G. prepared and tested the tutorial. V.A.H. has
split ratios will allow confirming the molecular ion, and the isotopic contributed with experimental work and with C.G-R. built the
pattern even for minority compounds. multimedia material. All authors contributed to write the article
and obtained the figures, A.G. and C.B. were responsible for the
design of the manuscript and the selection of content reported and
7.3. Case study
its supervision.
A set of 8 pool plasma samples were injected in a single Quad-

rupole gas chromatograph (GC-Q-MS) and another set of 8 pool Declaration of competing interest
plasma samples in a quadrupole time of flight (GC-QTOF-MS). In
the first one, conventional deconvolution was performed. In the The authors declare that they have no known competing
second one, SureMass was applied. With exception of the decon- financial interests or personal relationships that could have
volution algorithm, all the other parameters were fixed exactly the appeared to influence the work reported in this paper.
same. The spectra obtained after deconvolution were searched in
the same libraries, two of accurate and two of nominal mass and in Acknowledgment
the same order and, as for the deconvolution, the conditions were
set to the same values. Representative total ion chromatogram (TIC) This work was supported by Ministry of Science, Innovation and
of a QC plasma sample obtained with the high-resolution GC-MS Universities of Spain (MICINN) (Ref. RTI2018-095166-B-I00) co-
system is shown in Fig. 8 along with the list of metabolites found in funded by the European Regional Development Fund FEDER and
the profile depicted in Table 5. Comunidad de Madrid (B-2017/BMD-3751 “NOVELREN-CM” and
Fig. 9 illustrates the number of identified peaks with the low Ref. S2017/BMD3684 MOIR-2) and La Caixa Foundation (Ref. HR17-
resolution instrument: 143 and with the high resolution system: 00634).
19
Appendix A. Supplementary data [20] W.B. Dunn, I.D. Wilson, A.W. Nicholls, D. Broadhurst, The importance of
experimental design and QC samples in large-scale and MS-driven untargeted
metabolomic studies of humans, Bioanalysis 4 (2012) 2249e2264, https://
Supplementary data to this article can be found online at doi.org/10.4155/bio.12.204.
https://doi.org/10.1016/j.aca.2021.339043. [21] Z. He, Q. Luo, Z. Liu, L. Gong, Extensive evaluation of sample preparation
workflow for gas chromatography-mass spectrometry-based plasma metab-
olomics and its application in rheumatoid arthritis, Anal. Chim. Acta 1131
References (2020) 136e145, https://doi.org/10.1016/j.aca.2020.06.029.
[22] C. Sauerschnig, M. Doppler, C. Bueschl, R. Schuhmacher, Methanol generates
[1] G.J. Patti, O. Yanes, G. Siuzdak, Metabolomics: the apogee of the omics trilogy, numerous artifacts during sample extraction and storage of extracts in
Nat. Rev. Mol. Cell Biol. 13 (2012) 263e269, https://doi.org/10.1038/nrm3314. metabolomics research, Metabolites 8 (2017) 1, https://doi.org/10.3390/
[2] J.K. Nicholson, J.C. Lindon, Systems biology: metabonomics, Nature 455 (2008) metabo8010001.
1054e1056, https://doi.org/10.1038/4551054a. [23] A. Jiye, J. Trygg, J. Gullberg, A.I. Johansson, H. Antti, S.L. Marklund, T. Moritz,
[3] W.B. Dunn, D.I. Broadhurst, H.J. Atherton, R. Goodacre, J.L. Griffin, Systems Extraction and GC/MS analysis of the human blood plasma, Metabolome 77
level studies of mammalian metabolomes: the roles of mass spectrometry and (2005) 8086e8094.
nuclear magnetic resonance spectroscopy, Chem. Soc. Rev. 40 (2011) [24] J. Kim, J.K. Ahn, Y.E. Cheong, S.J. Lee, H.S. Cha, K.H. Kim, Systematic re-
387e426, https://doi.org/10.1039/b906712b. evaluation of the long-used standard protocol of urease-dependent metab-
[4] O. Fiehn, Metabolomics by gas chromatography-mass spectrometry: com- olome sample preparation, PloS One 15 (2020), https://doi.org/10.1371/
bined targeted and untargeted profiling, Curr. Protoc. Mol. Biol. 2016 (2016), journal.pone.0230072.
https://doi.org/10.1002/0471142727.mb3004s114. [25] V. Zaikin, J. Halket, A Handbook of Derivatives for Mass Spectrometry, IM
[5] T. Kind, G. Wohlgemuth, D.Y. Lee, Y. Lu, M. Palazoglu, S. Shahbaz, O. Fiehn, Publications, 2009.
FiehnLib: mass spectral and retention index libraries for metabolomics based [26] A. Chokkathukalam, D. Kim, M. Barrett, R. Breitling, D. Creek, Stable isotope-
on quadrupole and time-of-flight gas chromatography/mass spectrometry, labeling studies in metabolomics: new insights into structure and dynamics
Anal. Chem. 81 (2009) 10038e10048, https://doi.org/10.1021/ac9019522. of metabolic networks, Bioanalysis 6 (2014) 511e524, https://doi.org/
[6] W. Lu, X. Su, M.S. Klein, I.A. Lewis, O. Fiehn, J.D. Rabinowitz, Metabolite 10.4155/BIO.13.348.
measurement: pitfalls to avoid and practices to follow, Annu. Rev. Biochem. [27] A. Garcia, C. Barbas, Gas chromatography-mass spectrometry (GC-MS)-Based
86 (2017) 277e304, https://doi.org/10.1146/annurev-biochem-061516- metabolomics. Methods Mol. Biol., 2011, pp. 191e204, https://doi.org/
044952. 10.1007/978-1-61737-985-7_11.
[7] A. Mastrangelo, A. Ferrarini, F. Rey-Stolle, A. García, C. Barbas, From sample [28] A. Villasen ~ or, I. Garcia-Perez, A. Garcia, J.M. Posma, M. Ferna ndez-Lo pez,
treatment to biomarker discovery: a tutorial for untargeted metabolomics A.J. Nicholas, N. Modi, E. Holmes, C. Barbas, Breast milk metabolome charac-
based on GC-(EI)-Q-MS, Anal. Chim. Acta 900 (2015) 21e35, https://doi.org/ terization in a single-phase extraction, multiplatform analytical approach,
10.1016/j.aca.2015.10.001. Anal. Chem. 86 (2014) 8245e8252, https://doi.org/10.1021/ac501853d.
[8] J. Pezzatti, J. Boccard, S. Codesido, Y. Gagnebin, A. Joshi, D. Picard, V. Gonz
alez- [29] C. Gonzalez-Riano, G. Leo n-Espinosa, M. Regalado-Reyes, A. García, J. Defelipe,
Ruiz, S. Rudaz, Implementation of liquid chromatographyehigh resolution C. Barbas, Metabolomic study of hibernating Syrian hamster brains: in search
mass spectrometry methods for untargeted metabolomic analyses of biolog- of neuroprotective agents, J. Proteome Res. 18 (2019) 1175e1190, https://
ical samples: a tutorial, Anal. Chim. Acta (2020), https://doi.org/10.1016/ doi.org/10.1021/acs.jproteome.8b00816.
j.aca.2019.12.062. [30] S. Naz, A. García, C. Barbas, Multiplatform analytical methodology for meta-
[9] D. Stettin, R.X. Poulin, G. Pohnert, Metabolomics benefits from orbitrap bolic fingerprinting of lung tissue, Anal. Chem. 85 (2013) 10941e10948,
GCeMSdcomparison of low-and high-resolution GCeMS, Metabolites 10 https://doi.org/10.1021/ac402411n.
(2020), https://doi.org/10.3390/metabo10040143. [31] D. Gonza lez-Pen ~ a, D. Dudzik, A. García, B. Ancos, C. Barbas, C. Sanchez-Mor-
[10] D. Dudzik, C. Barbas-Bernardos, A. García, C. Barbas, Quality assurance pro- eno, Metabolomic fingerprinting in the comprehensive study of liver changes
cedures for mass spectrometry untargeted metabolomics. a review, associated with onion supplementation in hypercholesterolemic wistar rats,
J. Pharmaceut. Biomed. Anal. 147 (2018) 149e173, https://doi.org/10.1016/ Int. J. Mol. Sci. 18 (2017) 267, https://doi.org/10.3390/ijms18020267.
j.jpba.2017.07.044. [32] A. Navarrete, E.G. Armitage, M. Musteanu, A. García, A. Mastrangelo, R. Bujak,
[11] A.M. Evans, C. O'Donovan, M. Playdon, C. Beecher, R.D. Beger, J.A. Bowden, P.P. Lopez-Casas, M. Hidalgo, C. Barbas, Metabolomic evaluation of Mitomycin
D. Broadhurst, C.B. Clish, S. Dasari, W.B. Dunn, J.L. Griffin, T. Hartung, P.C. Hsu, C and rapamycin in a personalized treatment of pancreatic cancer, Pharmacol.
T. Huan, J. Jans, C.M. Jones, M. Kachman, A. Kleensang, M.R. Lewis, M.E. Monge, Res. Perspect. 2 (2014), https://doi.org/10.1002/prp2.67 n/a-n/a.
J.D. Mosley, E. Taylor, F. Tayyari, G. Theodoridis, F. Torta, B.K. Ubhi, [33] C. Gonzalez-Riano, M. Sanz-Rodríguez, J. Escudero-Ramirez, M.P. Lorenzo,
D. Vuckovic, Dissemination and analysis of the quality assurance (QA) and C. Barbas, B. Cubelos, A. Garcia, Target and untargeted GCeMS based metab-
quality control (QC) practices of LCeMS based untargeted metabolomics olomic study of mouse optic nerve and its potential in the study of neuro-
practitioners, Metabolomics 16 (2020), https://doi.org/10.1007/s11306-020- logical visual diseases, J. Pharmaceut. Biomed. Anal. 153 (2018) 44e56,
01728-5. https://doi.org/10.1016/j.jpba.2018.02.015.
[12] D. Broadhurst, R. Goodacre, S.N. Reinke, J. Kuligowski, I.D. Wilson, M.R. Lewis, [34] C. Gonzalez-Riano, S. Tapia-Gonza lez, A. García, A. Mun ~ oz, J. DeFelipe,
W.B. Dunn, Guidelines and considerations for the use of system suitability and C. Barbas, Metabolomics and neuroanatomical evaluation of post-mortem
quality control samples in mass spectrometry assays applied in untargeted changes in the hippocampus, Brain Struct. Funct. (2017), https://doi.org/
clinical metabolomic studies, Metabolomics 14 (2018) 72, https://doi.org/ 10.1007/s00429-017-1375-5.
10.1007/s11306-018-1367-3. [35] L.M. Pe rez, A. Bernal, B. de Lucas, N. San Martin, A. Mastrangelo, A. García,
[13] M.R. Viant, T.M.D. Ebbels, R.D. Beger, D.R. Ekman, D.J.T. Epps, H. Kamp, C. Barbas, B.G. G alvez, Altered metabolic and stemness capacity of adipose
P.E.G. Leonards, G.D. Loizou, J.I. MacRae, B. van Ravenzwaay, P. Rocca-Serra, tissue-derived stem cells from obese mouse and human, PloS One 10 (2015),
R.M. Salek, T. Walk, R.J.M. Weber, Use cases, best practice and reporting e0123397, https://doi.org/10.1371/journal.pone.0123397.
standards for metabolomics in regulatory toxicology, Nat. Commun. 10 [36] A. Binek, D. Rojo, J. Godzien, F.J. Rupe rez, V. Nunez, I. Jorge, M. Ricote,
(2019), https://doi.org/10.1038/s41467-019-10900-y. J. V
azquez, C. Barbas, Flow cytometry has a significant impact on the cellular
[14] Z. Pang, J. Chong, G. Zhou, D. de Lima Morais, L. Chang, M. Barrette, metabolome, J. Proteome Res. 18 (2019) 169e181, https://doi.org/10.1021/
C. Gauthier, P. Jacques, S. Li, J. Xia, MetaboAnalyst 5.0: narrowing the gap acs.jproteome.8b00472.
between raw spectra and functional insights, Nucleic Acids Res. 49 (2021), [37] G.A.B. Canuto, E.A. Castilho-Martins, M.F.M. Tavares, L. Rivas, C. Barbas,
Lo
A. pez-Gonz alvez, Multi-analytical platform metabolomic approach to
https://doi.org/10.1093/NAR/GKAB382.
[15] E.J. Want, P. Masson, F. Michopoulos, I.D. Wilson, G. Theodoridis, R.S. Plumb, study miltefosine mechanism of action and resistance in Leishmania, Anal.
J. Shockcor, N. Loftus, E. Holmes, J.K. Nicholson, Global metabolic profiling of Bioanal. Chem. 406 (2014) 3459e3476, https://doi.org/10.1007/s00216-014-
animal and human tissues via UPLC-MS, Nat. Protoc. 8 (2013) 17e32, https:// 7772-1.
doi.org/10.1038/nprot.2012.135. [38] D. Rojo, G.A.B. Canuto, E.A. Castilho-martins, F. Marina, A Multiplatform
[16] A. Wagner-Golbs, S. Neuber, B. Kamlage, N. Christiansen, B. Bethan, Metabolomic Approach to the Basis of Antimonial Action and Resistance in
U. Rennefahrt, P. Schatz, L. Lind, Effects of long-term storage at 80 C on the Leishmania Infantum, 2015, pp. 1e20, https://doi.org/10.1371/
human plasma metabolome, Metabolites 9 (2019) 99, https://doi.org/10.3390/ journal.pone.0130675.
metabo9050099. [39] F. Crepaldi, J.S. de Toledo, A.O. do Carmo, L. Ferreira Marques Machado,
[17] P. Yin, R. Lehmann, G. Xu, Effects of pre-analytical processes on blood samples D.D.V. de Brito, A.V. Serufo, A.P.M. Almeida, L.G. de Oliveira, T.Q.N. Ricotta, D.
used in metabolomics studies, Anal. Bioanal. Chem. 407 (2017) 4879e4892, de S. Moreira, S.M.F. Murta, A.B. Diniz, G.B. Menezes, A. Lo pez-Gonz alvez,
https://doi.org/10.1007/s00216-015-8565-x. C. Barbas, A.P. Fernandes, Mapping alterations induced by long-term axenic
[18] M. Haid, C. Muschet, S. Wahl, W. Ro € misch-Margl, C. Prehn, G. Mo € ller, cultivation of leishmania amazonensis promastigotes with a multiplatform
J. Adamski, Long-term stability of human plasma metabolites during storage metabolomic fingerprint approach, Front. Cell. Infect. Microbiol. 9 (2019),
https://doi.org/10.3389/fcimb.2019.00403.
at -80 c, J. Proteome Res. 17 (2018) 203e211, https://doi.org/10.1021/
acs.jproteome.7b00518. [40] E.V.C. Alves-Ferreira, J.S. Toledo, A.H.C. De Oliveira, T.R. Ferreira, P.C. Ruy,
[19] J. Godzien, V. Alonso-Herranz, C. Barbas, G. Armitage, Emily, Controlling the C.F. Pinzan, R.F. Santos, V. Boaventura, D. Rojo, A. Lopez-Gonza lvez, J.C. Rosa,
quality of metabolomics data : new strategies to get the best out of the QC C. Barbas, M. Barral-Netto, A. Barral, A.K. Cruz, Differential gene expression
sample, Metabolomics (2015) 518e528, https://doi.org/10.1007/s11306-014- and infection profiles of cutaneous and mucosal leishmania braziliensis iso-
0712-4. lates from the same patient, PLoS Neglected Trop. Dis. 9 (2015), https://
20
doi.org/10.1371/journal.pntd.0004018. Carolina Gonzalez-Riano obtained her PhD in Pharmacy at

[41] A.T. Faccio, F.J. Ruperez, N.S. Singh, S. Angulo, M.F.M. Tavares, M. Bernier, the Universidad CEU San Pablo, Madrid (Spain) under the
C. Barbas, I.W. Wainer, Stereochemical and structural effects of (2R,6R)- supervision of Prof. Coral Barbas and Prof. Antonia Garcia
hydroxynorketamine on the mitochondrial metabolome in PC-12 cells, Bio- in 2018. Currently, she is a postdoctoral researcher at the
chim. Biophys. Acta Gen. Subj. 1862 (2018) 1505e1515, https://doi.org/ Centre for Metabolomics and Bioanalysis (CEMBIO),
10.1016/j.bbagen.2018.03.008. Madrid (Spain). Her main field of interest involves targeted
[42] A. Mastrangelo, M.I. Panadero, L.M. Perez, B.G. Galvez, A. Garcia, C. Barbas, and untargeted metabolomics and lipidomics studies
F.J. Ruperez, New insight on obesity and adipose-derived stem cells using based on a multiplatform analytical approach (LC-MS,
comprehensive metabolomics, Biochem. J. 473 (2016) 2187e2203, https:// GC-MS, and CE-MS) to unveil the pathophysiological
doi.org/10.1042/BCJ20160241. mechanisms behind the development of different neuro-
[43] T. Davies, The new automated mass spectrometry deconvolution and identi- degenerative diseases. Her current research activity is
fication system (AMDIS), spectrosc, Eur 10/3 (1998) 24e27. also centered on applying a multiplatform untargeted
[44] Ł. Da˛ browski, Evaluation of a simplified method for GC/MS qualitative anal- metabolomics-based strategy to investigate tuberculosis.ResearcherID: K-2784-2015.
ysis of polycyclic aromatic hydrocarbons, polychlorinated biphenyls, and H-index: 6https://cembio.uspceu.es/author/carolina-
organic pesticides using PARADISe computer program, Molecules 25 (2020), gonzalez-riano/
https://doi.org/10.3390/MOLECULES25163727.
[45] F. Carnevale Neto, A. Pilon, D. Selegato, R. Freire, H. Gu, D. Raftery, N. Lopes,
I. Castro-Gamboa, Dereplication of natural products using GC-TOF mass PROF. Coral Barbas is currently Full Professor of
spectrometry: improved metabolite identification by spectral deconvolution Analytical Chemistry at Pharmacy Faculty, Uni-
ratio analysis, Front. Mol. Biosci. 3 (2016), https://doi.org/10.3389/ versidad San Pablo CEU, Madrid (Spain) and Director
FMOLB.2016.00059. for the “Centre for Metabolomics and Bioanalysis”
[46] Z. Lai, H. Tsugawa, G. Wohlgemuth, S. Mehta, M. Mueller, Y. Zheng, (CEMBIO). She is also Director for CEU International
A. Ogiwara, J. Meissen, M. Showalter, K. Takeuchi, T. Kind, P. Beal, M. Arita, School of Doctorate (CEINDO); Visiting Professor at
O. Fiehn, Identifying metabolites by integrating metabolome databases with Imperial College London. She is the author of more
mass spectrometry cheminformatics, Nat. Methods 15 (2018) 53e56, https:// than 300 papers, with current research interests in
doi.org/10.1038/NMETH.4512. all the steps of multiplatform metabolomics (GC-MS,
[47] A. Smirnov, W. Jia, D. Walker, D. Jones, X. Du, ADAP-GC 3.2: graphical software LC-MS and CE-MS) in all kind of biological samples.
tool for efficient spectral deconvolution of gas chromatography-high- Her awards include Angel Herrera Research and
resolution mass spectrometry metabolomics data, J. Proteome Res. 17 Teaching Award, the medal of Bialystok Medical Uni-
(2018) 470e478, https://doi.org/10.1021/ACS.JPROTEOME.7B00633. versity and recently she has received the award of
[48] X. Domingo-Almenara, J. Brezmes, M. Vinaixa, S. Samino, N. Ramirez, the Belgian Society of Pharmaceutical Sciences
M. Ramon-Krauel, C. Lerin, M. Díaz, L. Iba n
~ ez, X. Correig, A. Perera-Lluna, (BSPS 2018) and a Honoris causa doctorate in Bialystok Medical University Member
O. Yanes, eRah: a computational tool integrating spectral deconvolution and of different boards of international Committees and Editor for Journal of Pharmaceu-
alignment with quantification and identification of metabolites in GC/MS- tical and Biomedical Analysis.https://cembio.uspceu.es/members/
based metabolomics, Anal. Chem. 88 (2016) 9821e9829, https://doi.org/
10.1021/ACS.ANALCHEM.6B02927.
[49] A. Smirnov, Y. Qiu, W. Jia, D. Walker, D. Jones, X. Du, ADAP-GC 4.0: appli- Danuta Dudzik is an assistant professor at the Department of
cation of clustering-assisted multivariate curve resolution to spectral Biopharmacy and Pharmacodynamics at the Medical Univer-
deconvolution of gas chromatography-mass spectrometry metabolomics sk, Poland. She was a postdoctoral fellow and a
sity of Gdan
data, Anal. Chem. 91 (2019) 9069e9077, https://doi.org/10.1021/ research scientist in the Centre for Metabolomics and Bio-
ACS.ANALCHEM.9B01424. analysis University San Pablo CEU in Madrid (2010e2020).
[50] J. Stanstrup, C.D. Broeckling, R. Helmus, N. Hoffmann, E. Mathe , T. Naake, Her research is focused on perspectives for the application of
L. Nicolotti, K. Peters, J. Rainer, R.M. Salek, T. Schulze, E.L. Schymanski, untargeted metabolomics analyses in clinical practice, the use
M.A. Stravs, E.A. The venot, H. Treutler, R.J.M. Weber, E. Willighagen, of chromatographic techniques and mass spectrometry in
M. Witting, S. Neumann, The metaRbolomics toolbox in bioconductor and research on the pathogenesis of diseases, the development of
beyond, Metabolites 9 (2019), https://doi.org/10.3390/metabo9100200. new innovative diagnostic tools and systems biology. She co-
[51] W.B. Dunn, D. Broadhurst, P. Begley, E. Zelena, S. Francis-McIntyre, authored more than 38 original papers in peer-reviewed jour-
N. Anderson, M. Brown, J.D. Knowles, A. Halsall, J.N. Haselden, A.W. Nicholls, nals and 68 international conference communications
I.D. Wilson, D.B. Kell, R. Goodacre, Human Serum Metabolome (HUSERMET) including poster and oral presentations.
Consortium, Procedures for large-scale metabolic profiling of serum and
plasma using gas chromatography and liquid chromatography coupled to
mass spectrometry, Nat. Protoc. 6 (2011) 1060e1083, https://doi.org/10.1038/ David Rojo obtained his PhD in Pharmacy at Universidad
nprot.2011.335. CEU San Pablo, Madrid (Spain). Since then, he is a Post-
[52] J.A. Kirwan, D.I. Broadhurst, R.L. Davidson, M.R. Viant, Characterising and doctoral Researcher at CEMBIO (Madrid, Spain). He has
correcting batch variation in an automated direct infusion mass spectrometry developed his expertise in analytical chemistry with special
(DIMS) metabolomics workflow, Anal. Bioanal. Chem. 405 (2013) 5147e5157, interest on mass spectrometry. His current research activity
https://doi.org/10.1007/s00216-013-6856-7. is in the field of metabolomics; in particular, among other
[53] J. Kuligowski, A. Sa
nchez-Illana, D. Sanju an-Herr aez, M. Vento, G. Quintas, topics and projects, he is focused on the metabolomics of
Intra-batch effect correction in liquid chromatography-mass spectrometry gut microbiota and the data integration for systems
using quality control samples and support vector regression (QC-SVRC), An- biology. He has directed several Master projects.
alyst 140 (2015) 7810e7817, https://doi.org/10.1039/C5AN01638J. Throughout his career, he has been recognized for the ac-
[54] R.A. van den Berg, H.C.J. Hoefsloot, J.A. Westerhuis, A.K. Smilde, M.J. van der ademic excellence of his research through a series of
Werf, Centering, scaling, and transformations: improving the biological in- awards and state scholarships, and twice, one of his articles
formation content of metabolomics data, BMC Genom. 7 (2006) 142, https:// has been selected for the cover page. He authored more
doi.org/10.1186/1471-2164-7-142. than 40 scientific articles in journals included in JCR and 3
book chapters. Up to date, his research has received more than 1500 cites.
PROF. Antonia García is currently Full Professor of Analyt-

ical Chemistry at Pharmacy Faculty, Universidad San Pablo Fernanda Rey-Stolle is senior lecturer and researcher at San
CEU, Madrid (Spain) and cofounder members of the Centre Pablo CEU University since 1994. Since 2013 she is a mem-
for Metabolomics and Bioanalysis of San Pablo CEU Uni- ber of CEMBIO (Centre of Excellence in Metabolomics and
versity (Madrid, Spain) in 2008. Her research has been Bioanalysis), focusing her research on the application of
focused on development and validation of chromato- the GC/MS technique to metabolomic studies. She has
graphic methods for metabolomics based on mass spec- participated in 16 Projects and/or Research Contracts, in 2
trometry multiplatforms (LC, CE and GC) in the study of of them as PI. She is coauthor of 43 articles and has pre-
several diseases such as cancer, diabetes, obesity, infec- sented more than 50 oral communications and posters in
tious, lung and cardiovascular diseases. 8 supervised the- relevant International Congresses. She has been a member
ses. National and international grants as PI. Journal of of the scientific/organizing committee of 9 International
Pharmaceutical and Biomedical Analysis Editorial Board member. More than 110 paper and National congresses.https://cembio.uspceu.es/?
reviewed for different international journals.Researcher ID: C-4296-2008. Pub papers. author¼13
(WoS Core): 90.https://cembio.uspceu.es/?author¼4
21
Miguel Ferna ndez-García obtained his PhD in Pharmacy at Vanesa Alonso is a Senior Technician since 2004, and Mass
San Pablo CEU University (Madrid, Spain) in 2021. He Spectrometry Technician at Universidad San Pablo CEU
joined the Centre for Metabolomics and Bioanalysis since 2005. She is currently Lab Manager at CEMBIO lab-
(CEMBIO) of San Pablo CEU University in 2016. Currently, oratory, where she has acquired and refined her expertise
he is a Postdoctoral Researcher at CEMBIO with a special in LC-IT-MS, LC-QTOF-MS, LC-QqQ-MS, CE-TOF-MS, GC-Q-
focus in biochemistry and bioinformatics. His research MS, GC-QTOF-MS and sample treatment in bio-
focuses on the application of hyphenated, mass analysis.https://cembio.uspceu.es/?author¼15
spectrometry-based analytical techniques and bio-
informatic pipelines to provide insights into the pathogen-
esis of human infectious diseases caused
by Mycobacterium tuberculosis, Haemophilus
influenzae, Leishmania donovani, Babesia divergens and
Aspergillus fumigatus.Researcher ID: A-6895-2017.https://
cembio.uspceu.es/?author¼24
22

1 s2.0 S0003267021008692 Main

Uploaded by

Copyright:

Available Formats

1 s2.0 S0003267021008692 Main

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

1 s2.0 S0003267021008692 Main

Uploaded by

Copyright:

Available Formats

Analytica Chimica Acta xxx (xxxx) xxx

Contents lists available at ScienceDirect

Analytica Chimica Acta

Low and high resolution gas chromatography-mass spectrometry for

Ready-to-use protocols for untar-

Abbreviations GC-QTOF-MS Gas Chromatography-Time-of-Flight Mass

1. Introduction compounds, typically <1500 Da) in a given sample obtained from a

FEATURES GC-Q-MS GC-QTOF-MS

Metabolomic approach (targeted, untargeted): absolute and Yes Yes

Fig. 1. Experimental design of the GC-MS metabolomic experiment.

Plasma/Serum, Bronchoalveolar lavage ﬂuid [27] Sample collection

3. Evaporation TIMING ~ 2 h 3. Evaporation TIMING ~ 2 h

SAMPLE TYPE SAMPLE METHOXYMATION SILYLATION SET UP FOR

TIMING ~0.5 h þ 16 h TIMING ~ 2 h TIMING ~

Add 350 mL pre-cooled methanol (-20 C) with 25 ppm of d31-

3. Evaporation TIMING ~ 2 h 1. Homogenization and sample preprocessing TIMING ~ 0.5 h

Helium (99.9999%) carrier gas. Flow rate: 1 mL/min.

4.1. Instrument set-up 4.3. Calibration of the instrument

Sample type Characterization Nr of injections Time (approx.)

System suitability IS Methyl stearate (IS) 10 mg/L in n-heptane 5 200 min

Fig. 3. GC-MS analysis checklist for low- and high-resolution system.

Practical troubleshootings - The most common problems

Problem Possible causes Solution

6.2. Data ﬁltering 7.1. Curation of tables

Transformation can be applied in metabolite abundances from

number Name RT (min) number Name RT (min)

1 Boric acid, 3TMS 5.74 79 Pyrophosphate, 4TMS 14.73

number Name RT (min) number Name RT (min)

75 Triethanolamine, 3TMS 14.37 153 Glycerol monostearate, 2TMS 24.87

A set of 8 pool plasma samples were injected in a single Quad-

doi.org/10.1371/journal.pntd.0004018. Carolina Gonzalez-Riano obtained her PhD in Pharmacy at

PROF. Antonia García is currently Full Professor of Analyt-

You might also like