Automating data analysis for two-dimensional gas chromatography/time-of-flight mass spectrometry non-targeted analysis of comparative samples

Region of interest selection in heterogeneous digital image: Wine age prediction by comprehensive two-dimensional gas chromatography

This study integrates genetic algorithm (GA) with partial least squares regression (PLSR) and various variable selection methods to identify impactful regions of interest (ROI) in heterogeneous 2D chromatogram images for predicting wine age. As wine quality and aroma evolve over time, transitioning from youthful fruitiness to mature, complex flavors, which leads to alterations in the composition of essential aroma-contributing compounds. Chromatograms are segmented into subimages, and the GA-PLSR algorithm optimizes combinations based on grayscale, red-green-blue (RGB), and hue-saturation-value (HSV) histograms. The selected subimage histograms are further refined through interval selection, highlighting the compounds with the most significant influence on wine aging. Experimental validation involving 38 wine samples demonstrates the effectiveness of this approach. Cross-validation reduces the PLS model error from 2.8 to 2.4 years within a 10 × 10 subset, and during prediction, the error decreases from 2.5 to 2.3 years. The study presents a novel approach utilizing the selection of ROI for efficient processing of 2D chromatograms focusing on predicting wine age.

A non-target geographical origin screening of botrytized wines through comprehensive two-dimensional gas chromatography coupled with high-resolution mass spectrometry

One of the most effective methods for gaining insight into the composition of trace-level volatile organic characteristics of wine products is through the use of a comprehensive two-dimensional gas chromatography-high resolution mass spectrometry (GC × GC-HRMS) technique. The vast amount of data generated by this method, however, can often be overwhelming requiring exhaustive and time-consuming analysis to identify significant statistical characteristics. The use of advanced chemometric software can achieve the same or even higher efficiency. This study aimed to identify differences based on geographical locations by analyzing the volatile organic compounds in the composition of botrytized wines from Slovakia, Hungary, France, and Austria. The volatile organic compounds were extracted by solid-phase microextraction and analyzed using GC × GC-HRMS. The data obtained from the analysis underwent Fisher-ratio (F-ratio) tile-based analysis to identify statistically significant differences. Principal component analysis demonstrated a significant distinction between wine samples based on geographical location, using only 10 statistically significant features with the highest F-ratio. In the samples, the following compounds were analyzed: methyl-octadecanoate, 2-cyanophenyl-β-phenylpropionate, α-ionone, n-octanoic acid, 1,2-dihydro-1,1,6-trimethyl-naphthalene, methyl-hexadecanoate, ethyl-pentadecanoate, ethyl-decanoate, and γ-nonalactone. These, all play an important role in cluster pattern observed on principal component analysis results. Additionally, hierarchical cluster analysis confirmed this.

Aroma Compounds in Essential Oils: Analyzing Chemical Composition Using Two-Dimensional Gas Chromatography-High Resolution Time-of-Flight Mass Spectrometry Combined with Chemometrics

Analyzing essential oils is a challenging task for chemists because their composition can vary depending on various factors. The separation potential of volatile compounds using enantioselective two-dimensional gas chromatography coupled with high-resolution time-of-flight mass spectrometry (GC×GC-HRTOF-MS) with three different stationary phases in the first dimension was evaluated to classify different types of rose essential oils. The results showed that selecting only ten specific compounds was enough for efficient sample classification instead of the initial 100 compounds. The study also investigated the separation efficiencies of three stationary phases in the first dimension: Chirasil-Dex, MEGA-DEX DET-β, and Rt-βDEXsp. Chirasil-Dex had the largest separation factor and separation space, ranging from 47.35% to 56.38%, while Rt-βDEXsp had the smallest, ranging from 23.36% to 26.21%. MEGA-DEX DET-β and Chirasil-Dex allowed group-type separation based on factors such as polarity, H-bonding ability, and polarizability, whereas group-type separation with Rt-βDEXsp was almost imperceptible. The modulation period was 6 s with Chirasil-Dex and 8 s with the other two set-ups. Overall, the study showed that analyzing essential oils using GC×GC-HRTOF-MS with a specific selection of compounds and stationary phase can be effective in classifying different oil types.

Biological studies with comprehensive 2D-GC-HRMS screening: Exploring the human sweat volatilome

A key issue in GCxGC-HRMS data analysis is how to approach large-sample studies in an efficient and comprehensive way. We have developed a semi-automated data-driven workflow from identification to suspect screening, which allows highly selective monitoring of each identified chemical in a large-sample dataset. The example dataset used to illustrate the potential of the approach consisted of human sweat samples from 40 participants, including field blanks (80 samples). These samples have been collected in a Horizon 2020 project to investigate the capacity of body odour to communicate emotion and influence social behaviour. We used dynamic headspace extraction, which allows comprehensive extraction with high preconcentration capability, and has to date only been used for a few biological applications. We were able to detect a set of 326 compounds from a diverse range of chemical classes (278 identified compounds, 39 class unknowns, and 9 true unknowns). Unlike partitioning-based extraction methods, the developed method detects semi-polar (log P < 2) nitrogen and oxygen-containing compounds. However, it is unable to detect certain acids due to the pH conditions of unmodified sweat samples. We believe that our framework will open up the possibility of efficiently using GCxGC-HRMS for large-sample studies in a wide range of applications such as biological and environmental studies.

Screening organic contaminants in soil by two-dimensional gas chromatography high-resolution time-of-flight mass spectrometry: A non-target analysis strategy and contaminated area case study

Thousands of organic substances that are used in industrial applications ultimately enter the soil and may negatively affect human health. Limited numbers of target pollutants are usually monitored in environmental media because of analytical limitations. In this study, a non-target screening method for quickly analyzing multiple soil samples from a contaminated area (a chemical industry park) by two-dimensional gas chromatography high-resolution time-of-flight mass spectrometry was developed. The types of compounds present in the soil samples were preliminarily analyzed through data simplification and visual assessment. A total of 81 organic compounds with detection frequencies ≥40% in the samples from the chemical industry park were selected for identification, including 38 PAHs, 26 oxygenated organic compounds, eight N-containing compounds, and nine other compounds. Potential sources of the organic compounds in the industrial park were investigated. Some pharmaceutical and organic synthetic intermediates in the soil were affected by nearby chemical plants. After assessing the relative abundances and detection frequencies, 36 pollutants that may pose potential risks to the environment were preliminarily identified. The results of the study were helpful for assessing environmental risks around Yangkou industrial park and they will be helpful when assessing risks in other contaminated areas.

Automated high confidence compound identification of electron ionization mass spectra for nontargeted analysis

Nontargeted analysis based on mass spectrometry is a rising practice in environmental monitoring for identifying contaminants of emerging concern. Nontargeted analysis performed using comprehensive two-dimensional gas chromatography coupled with time-of-flight mass spectrometry (GC×GC/TOF-MS) generates large numbers of possible analytes. Moreover, the default spectral library similarity score-based search algorithm used by LECO® ChromaTOF® does not ensure that high similarity scores result in correct library matches. Therefore, an additional manual screening is necessary, but leads to human errors especially when dealing with large amounts of data. To improve the speed and accuracy of the chemical identification, we developed CINeMA.py (Classification Is Never Manual Again). This programming suite automates GC×GC/TOF-MS data interpretation by determining the confidence of a match between the observed analyte mass spectrum and the LECO® ChromaTOF® software generated library hit from the NIST Electron Ionization Mass Spectral (NIST EI-MS) library. Our script allows the user to evaluate the confidence of the match using an algorithmic method that mimics the manual curation process and two different machine learning approaches (neural networks and random forest). The script allows the user to adjust various parameters (e.g., similarity threshold) and study their effects on prediction accuracy. To test CINeMA.py, we used data from two different environmental contaminant studies: an EPA study on household dust and a study on stormwater runoff. Using a reference set based on the analysis performed by highly trained users of the ChromaTOF and GC×GC/TOF-MS systems, the random forest model had the highest prediction accuracies of 86% and 83% on the EPA and Stormwater data sets, respectively. The algorithmic approach had the second-best prediction accuracy (82% and 79%), while the neural network accuracy had the lowest (63% and 67%). All the approaches required less than 1 min to classify 986 observed analytes, whereas manual data analysis required hours or days to complete. Our methods were also able to detect high confidence matches missed during the manual review. Overall, CINeMA.py provides users with a powerful suite of tools that should significantly speed-up data analysis while reducing the possibilities of manual errors and discrepancies among users, and can be applicable to other GC/EI-MS instrument based nontargeted analysis.

Automated Screening and Filtering Scripts for GC×GC-TOFMS Metabolomics Data

Comprehensive two-dimensional gas chromatography mass spectrometry (GC×GC-MS) is a powerful tool for the analysis of complex mixtures, and it is ideally suited to discovery studies where the entire sample is potentially of interest. Unfortunately, when unit mass resolution mass spectrometers are used, many detected compounds have spectra that do not match well with libraries. This could be due to the compound not being in the library, or the compound having a weak/nonexistent molecular ion cluster. While high-speed, high-resolution mass spectrometers, or ion sources with softer ionization than 70 eV electron impact (EI) may help with some of this, many GC×GC systems presently in use employ low-resolution mass spectrometers and 70 eV EI ionization. Scripting tools that apply filters to GC×GC-TOFMS data based on logical operations applied to spectral and/or retention data have been used previously for environmental and petroleum samples. This approach rapidly filters GC×GC-TOFMS peak tables (or raw data) and is available in software from multiple vendors. In this work, we present a series of scripts that have been developed to rapidly classify major groups of compounds that are of relevance to metabolomics studies including: fatty acid methyl esters, free fatty acids, aldehydes, alcohols, ketones, amino acids, and carbohydrates.

Automating and Extending Comprehensive Two-Dimensional Gas Chromatography Data Processing by Interfacing Open-Source and Commercial Software

Comprehensive two-dimensional gas chromatography (GC×GC) is a powerful analytical tool for both nontargeted and targeted analyses. However, there is a need for more integrated workflows for processing and managing the resultant high-complexity datasets. End-to-end workflows for processing GC×GC data are challenging and often require multiple tools or software to process a single dataset. We describe a new approach, which uses an existing underutilized interface within commercial software to integrate free and open-source/external scripts and tools, tailoring the workflow to the needs of the individual researcher within a single software environment. To demonstrate the concept, the interface was successfully used to complete a first-pass alignment on a large-scale GC×GC metabolomics dataset. The analysis was performed by interfacing bespoke and published external algorithms within a commercial software environment to automatically correct the variation in retention times captured by a routine reference standard. Variation in ¹t_R and ²t_R was reduced on average from 8 and 16% CV prealignment to less than 1 and 2% post alignment, respectively. The interface enables automation and creation of new functions and increases the interconnectivity between chemometric tools, providing a window for integrating data-processing software with larger informatics-based data management platforms.

Application of thermal desorption methods for airborne polycyclic aromatic hydrocarbon measurement: A critical review

Thermal desorption (TD) is a universal solvent-free pre-concentration technique. It is often used to pre-concentrate semi-volatile and volatile organic compounds in various sample types. Polycyclic aromatic hydrocarbons (PAHs) are widespread contaminants from incomplete combustion of organic matter and fossil fuel, which have carcinogenic effects on human health. Conventional methods for determining PAHs, represented by solvent extraction, are gradually being replaced by solvent-free methods, typically the TD technique, because of TD's many advantages, including time savings and environmentally friendly treatment. This work presents an extensive review of the universal methods used to determine PAHs in the atmosphere based on the TD technique. The methods currently used for collection and detection of both gas- and particle-phase PAHs in the air are critically reviewed. In addition, the operating parameters of the TD unit are summarized and discussed. The design shortcomings of existing studies and the problems that researchers should address are presented, and promising alternatives are suggested. This paper also discusses important parameters, such as reproducibility and limit of detection, that form a crucial part of quality assurance. Finally, the limitations and the future prospects of the TD technique for use in airborne PAH analyses are addressed. This is the first review of the latest developments of the TD technique for analysis of PAHs and their derivatives in the atmosphere.

A Data-Challenge Case Study of Analyte Detection and Identification with Comprehensive Two-Dimensional Gas Chromatography with Mass Spectrometry (GC×GC-MS)

This case study describes data analysis of a chromatogram distributed for the 2019 GC×GC Data Challenge for the Tenth Multidimensional Chromatography Workshop (Liege, Belgium). The chromatogram resulted from chemical analysis of a terpene-standards sample by comprehensive two-dimensional chromatography with mass spectrometry (GC×GC-MS). First, several aspects of the data quality are assessed, including detector saturation and oscillation, and operations to prepare the data for analyte detection and identification are described, including phase roll for modulation-cycle alignment and baseline correction to account for the non-zero detector baseline. Then, the case study presents operations for analyte detection with filtering, a new method to flag false detections, interactive review to confirm detected peaks, and ion-peaks detection to reveal peaks that are obscured by noise or coelution. Finally, the case study describes analyte identification including mass-spectral library search with a new method for optimizing spectra extraction, retention-index calibration from preliminary identifications, and expression-based identification checks. Processing of the first 40 min of data detected 144 analytes, 21 of which have at least one percent response, plus an additional 20 trace and/or coeluted analytes.

Formation of Polycyclic Aromatic Hydrocarbon Oxidation Products in α-Pinene Secondary Organic Aerosol Particles Formed through Ozonolysis

Accurate long-range atmospheric transport (LRAT) modeling of polycyclic aromatic hydrocarbons (PAHs) and PAH oxidation products (PAH-OPs) in secondary organic aerosol (SOA) particles relies on the known chemical composition of the particles. Four PAHs, phenanthrene (PHE), dibenzothiophene (DBT), pyrene (PYR), and benz(a)anthracene (BaA), were studied individually to identify and quantify PAH-OPs produced and incorporated into SOA particles formed by ozonolysis of α-pinene in the presence of PAH vapor. SOA particles were characterized using real-time in situ instrumentation, and collected on quartz fiber filters for offline analysis of PAHs and PAH-OPs. PAH-OPs were measured in all PAH experiments at equal or greater concentrations than the individual PAHs they were produced from. The total mass of PAH and PAH-OPs, relative to the total SOA mass, varied for different experiments on individual parent PAHs: PHE and 6 quantified PHE-OPs (3.0%), DBT and dibenzothiophene sulfone (4.9%), PYR and 3 quantified PYR-OPs (3.1%), and BaA and benz(a)anthracene-7,12-dione (0.26%). Further exposure of PAH-SOA to ozone generally increased the concentration ratio of PAH-OPs to PAH, suggesting longer atmospheric lifetimes for PAH-OPs, relative to PAHs. These data indicate that PAH-OPs are formed during SOA particle formation and growth.

Nontargeted Detection Methods for Food Safety and Integrity

Nontargeted workflows for chemical hazard analyses are highly desirable in the food safety and integrity fields to ensure human health. Two different analytical strategies, nontargeted metabolomics and chemical database filtering, can be used to screen unknown contaminants in food matrices. Sufficient mass and chromatographic resolutions are necessary for the detection of compounds and subsequent componentization and interpretation of candidate ions. Analytical chemistry–based technologies, including gas chromatography–mass spectrometry (GC-MS), liquid chromatography–mass spectrometry (LC-MS), nuclear magnetic resonance (NMR), and capillary electrophoresis–mass spectrometry (CE-MS), combined with chemometrics analysis are being used to generate molecular formulas of compounds of interest. The construction of a chemical database plays a crucial role in nontargeted detection. This review provides an overview of the current sample preparation, analytical chemistry–based techniques, and data analysis as well as the limitations and challenges of nontargeted detection methods for analyzing complex food matrices. Improvements in sample preparation and analytical platforms may enhance the relevance of food authenticity, quality, and safety.

Evaluating Computational and Structural Approaches to Predict Transformation Products of Polycyclic Aromatic Hydrocarbons

Polycyclic aromatic hydrocarbons (PAHs) undergo transformation reactions with atmospheric photochemical oxidants, such as hydroxyl radicals (OH•), nitrogen oxides (NOx), and ozone (O3). The most common PAH-transformation products (PAH-TPs) are nitrated, oxygenated, and hydroxylated PAHs (NPAHs, OPAHs, and OHPAHs, respectively), some of which are known to pose potential human health concerns. We sampled four theoretical approaches for predicting the location of reactive sites on PAHs (i.e., the carbon where atmospheric oxidants attack), and hence the chemoselectivity of the PAHs. All computed results are based on density functional theory (B3LYP/6-31G(d) optimized structures and energies). The four approaches are (1) Clar's prediction of aromatic resonance structures, (2) thermodynamic stability of all OHPAH adduct intermediates, (3) computed atomic charges (Natural Bond order, ChelpG, and Mulliken) at each carbon on the PAH, and (4) average local ionization energy (ALIE) at atom or bond sites. To evaluate the accuracy of these approaches, the predicted PAH-TPs were compared to published laboratory observations of major NPAH, OPAH, and OHPAH products in both gas and particle phases. We found that the Clar's resonance structures were able to predict the least stable rings on the PAHs but did not offer insights in terms of which individual carbon is most reactive. The OHPAH adduct thermodynamics and the ALIE approaches were the most accurate when compared to laboratory data, showing great potential for predicting the formation of previously unstudied PAH-TPs that are likely to form in the atmosphere.