Illuminating the Dark Multi-Omes: Advancing Molecular Characterization with IM-MS

At the molecular level, understanding the human body involves wrestling with vast amounts of data that defy identification, a challenge central to metabolomics. The metabolome, which is the entire set of small molecules in a biological system, contains peptides, lipids, and carbohydrates, including signatures that do not match any previously identified molecular structure. This unidentified material is often referred to as the dark metabolome or, more generally, the dark multi-ome. Metabolomics, much like genomics, transcriptomics, and proteomics, aims to understand health and disease, serving as a powerful tool for hypothesis generation in biological research. The most common analytical platform for metabolomics is liquid chromatography–mass spectrometry (LC–MS), which generates large datasets requiring sophisticated software and bioinformatic tools for processing and interpretation.

However, the sheer volume and complexity of data pose significant challenges, including difficulties in experimental optimization, extraction of low-abundance but biologically meaningful features, accurate quantification, cleanup and deconvolution of spectral complexity, and confirmation of metabolite identities (Figure 1) (1). These “big data challenges” are often cited as one of the limiting factors in the widespread application of metabolomics (2). This article delves into how advanced mass spectrometry techniques, particularly those incorporating ion mobility spectrometry (IMS), are addressing these fundamental hurdles, paving the way for more comprehensive and reliable insights into the dark “omics.”

Ion Mobility Advancements and Utility for Metabolomics

Ion mobility integrated with mass spectrometry (IM-MS) has emerged as a transformative technology in metabolomics research, providing multidimensional characterization of metabolites that was previously unachievable. Historically, while mass spectrometry alone offered advancements in detecting and quantifying small molecules, it faced limitations in resolving isomeric compounds and chemical interferences from analytes of interest. The integration of gas-phase ion mobility separations, operating on a millisecond timescale, with existing chromatographic methods such as LC and time-of-flight mass spectrometry (TOF-MS), addresses these challenges by offering an additional dimension of separation. This combination, often referred to as LC–IM-MS, significantly enhances selectivity and sensitivity, enabling the resolution of exogenous isomeric compounds and improved molecular annotation through the measurement of a unique molecular descriptor: collision cross-section (CCS). CCS values are notably reproducible across instruments and laboratories, making them ideal for standardization and for improving the accuracy of metabolite identification (3).

The field of IM-MS has seen rapid development, with several distinct technologies contributing to its growing impact on metabolomics (4):

• Drift Tube Ion Mobility Spectrometry (DTIMS): This foundational technique operates by propelling ions through a buffer gas under a uniform electric field. It provides the most accurate means for measuring CCS based on first principles, with smaller ions traversing the drift tube more quickly.
• Traveling Wave Ion Mobility Spectrometry (TWIMS): In contrast, TWIMS employs a dynamic, migrating electrical potential (a “traveling wave”) to move ions through the buffer gas. Similar to DTIMS, smaller ions experience fewer collisions and arrive faster.
• Trapped Ion Mobility Spectrometry (TIMS): This technique uses an analyzer cell to trap ions between a gas flow and a stopping potential. Ions are then selectively released by gradually decreasing the stopping potential, with molecules possessing the largest CCS values arriving first.
• Cyclic Ion Mobility (cIM): A significant advancement in TWIMS, cIM incorporates a closed-loop path that allows ions to undergo multiple passes, thereby iteratively enhancing resolution. This multi-pass capability also enables advanced IMSn experiments, including mobility selection, activation/fragmentation, and separation of the reaction products.
• Structures for Lossless Ion Manipulations (SLIM): SLIM technology uses a printed circuit board architecture to create extended, serpentine ion trajectories, and uses traveling wave-based separations. SLIM-based IMS offers unique capabilities that position it as a prospective technology for broad clinical and bioanalytical applications. It provides broadband and high-resolution mobility separations, with commercial instruments achieving resolving powers of 200–300 Rp without the need for post-processing or narrowed measurement windows (5). This contrasts with conventional technologies that typically max out natively at 40–60 Rp. The long mobility path length is a primary driver of this enhanced resolving power. Furthermore, SLIM is compatible with LC separations, allowing for sequential multidimensional analysis without sacrificing throughput, as IM separations occur on a faster timescale than LC. Its ability to generate high-quality data across a broad mass and mobility range is supported by advanced acquisition systems, handling high ion currents in millisecond-wide peaks and offering wide molecular coverage for multi-omics (6). This allows for the confident detection and characterization of diverse molecular classes, including low-abundance isomers, without the need for specific “zooming” into narrow mass-to-charge (m/z) windows for enhanced resolution, providing more selective and unambiguous identification.

Collectively, the rapid evolution and diversification of IM-MS platforms represent an inflection point for metabolomics research. These innovations directly address longstanding challenges in distinguishing chemically similar species and isomeric compounds, which are often intractable by mass spectrometry alone. By providing an additional dimension of separation based on molecular size and shape, alongside reproducible CCS values, IM-MS significantly enhances analytical selectivity and confidence in molecular identification, especially at low concentrations. The advent of high-resolution IM (HRIM) capabilities, exemplified by technologies like SLIM, will provide unprecedented peak capacity and enable the resolution of analytes with subtle structural differences. This collective enhancement in analytical power is pivotal for achieving precise quantitative and confident qualitative analyses in complex biological matrices. The following sections will highlight how these advanced IM-MS capabilities are addressing critical analytical challenges, from enhancing isomer separation and identification to improving quantitative performance for broad clinical and bioanalytical applications.

Isomeric Ambiguity

One of the most significant challenges in metabolomics research is isomeric ambiguity, where a large proportion of metabolites share the same molecular mass and chemical formula, making them indistinguishable by mass spectrometry alone. Studies analyzing major metabolite databases indicate that only 37–45% of metabolites possess unique masses, with the remainder existing as isobaric or isomeric species (7). Hence, traditional workflows based only on MS1 fail to annotate, measure, and report the majority of metabolites present in a sample.

LC has served as the primary method for separating isomeric metabolites, with different chromatographic phases offering specific advantages; for example, reversed-phase high performance LC is effective for separating positional isomers through hydrophobic interactions. However, there is not a single LC workflow that is optimized to separate the entire metabolome. Notably, IMS can be combined with LC to achieve improved overall analytical resolution. IMS introduces an orthogonal dimension of separation based on the CCS of ions, which is determined by their size, shape, and charge in the gas phase. The combination of IM-MS, LC, and tandem MS (LC–IM-MS/MS) substantially improves the resolution of coeluting species and isomers that would otherwise remain unresolved by LC–MS alone.

The metabolomics community is actively working toward creating comprehensive CCS databases and standardizing chromatographic conditions to further leverage these advancements. Ultimately, the integration of IMS with mass spectrometry allows for a multidimensional analysis, where retention time, m/z, and CCS combine to provide a more robust and confident identification of otherwise ambiguous isomeric compounds.

Low-abundance Detection

The detection of low-abundance metabolites poses a formidable analytical challenge in metabolomics. The human metabolome encompasses over 100,000 endogenous metabolites and 1,000,000 exogenous metabolites, spanning an enormous dynamic concentration (8,9). This wide dynamic range means that highly abundant metabolites can often mask the signals from trace-level compounds, even though these low-abundance metabolites may hold significant biological importance. Furthermore, low-abundance metabolites frequently exhibit poor signal-to-noise ratios and greater imprecision between analytical runs, complicating and potentially compromising data quality. Accurate identification of these elusive compounds is also hampered by the frequent unavailability of commercial analytical standards, particularly for novel or uncommon metabolites.

To address these challenges, several advanced analytical and computational approaches are being developed. Mass spectrometry optimization includes the use of ultra-sensitive targeted methods, such as trapping-micro-LC coupled with narrow-window-isolation selected-reaction monitoring, which have achieved impressive limits of quantification down to picogram-per-milliliter levels for certain compounds (10). Derivatization, solid-phase extraction and cleanup, and pre-concentration strategies are regularly used to overcome sensitivity limitations with varying degrees of success and added complexity (11).

IM-MS represents a significant leap forward in enhancing sensitivity and coverage for low-abundance analytes. IM-MS, particularly with rapidly advancing hardware, allows for acquisition rates exceeding 500 Hz, enabling the generation of fragmentation data for more precursor ions including low-intensity features. The parallel accumulation–serial fragmentation (PASEF) principle, implemented on TIMS-quadrupole time-of-flight (QTOF) mass spectrometers, is particularly impactful. PASEF accumulates ions in parallel and releases them sequentially, leading to a signal amplification of more than an order of magnitude compared to continuous acquisition without TIMS, increasing the signal-to-noise ratio. This capability allows PASEF to increase throughput without compromising sensitivity, making it highly attractive for sensitive lipidomics, even from minimal sample amounts, achieving attomole-range detection limits for certain lipid classes.

Whereas traditional data-dependent acquisition (DDA) often falls short in fragmenting low-abundance compounds due to its bias toward more abundant ions, full-scan mode (acquisition of only MS1 data) generally captures the largest number of metabolic features, including more low-abundance signals, because 100% of available instrument time is dedicated to MS1 data acquisition. However, it is important to note that while IMS excels at cleaning spectra and improving selectivity, analyses with some forms of IM-MS can lead to a loss of lower intensity metabolites compared to methods without IM, highlighting a balance that needs to be considered based on experimental goals (12).

Chimeric MS/MS Spectra

Chimeric MS/MS spectra represent a substantial challenge in mass spectrometry-based omics research. These spectra contain fragments originating from multiple precursor ions that were simultaneously fragmented. The presence of chimeric spectra can artificially inflate spectral match scores between unrelated compounds, leading to incorrect clustering in molecular networks and compromising the biological interpretation of metabolomic data (13). Ultimately, they hinder accurate compound identification because they do not match the clean, single-compound reference spectra typically found in metabolomic databases.

The widespread adoption of data-independent acquisition (DIA) workflows has exacerbated this issue. DIA, unlike DDA, fragments all analyte ions at a given time simultaneously rather than selectively targeting specific ions. While this unbiased approach offers comprehensive data collection, it results in highly complex, multiplexed MS/MS spectra, making metabolite identification difficult because of the dissociation of the direct link between precursor and fragment ions.

To overcome the challenges posed by chimeric spectra, sophisticated computational deconvolution solutions are essential to improve identification rates by database-assisted MS/MS deconvolution and process data from various instrument vendors for both DDA and DIA workflows.

IM-MS further contributes to addressing chimeric spectra by adding an extra dimension of separation. IM-MS decreases spectral complexity and improves precursor specificity by separating coeluting species in the gas phase based on their CCS prior to fragmentation (Figure 2). In IM-MS-enabled DIA, coeluting precursor ions can be separated in the drift time dimension before fragmentation, allowing fragments that do not match the precursor ion’s drift time to be filtered out, thereby yielding cleaner spectra. The metabolomics community is also working toward standardized approaches for handling complex spectra, including common file formats, quality metrics, and validation procedures across different analytical platforms and software tools, which
will enhance the reliability and coverage of metabolomic analyses (14).

In-source Fragmentation Artifacts

In-source fragmentation (ISF) refers to the unintended fragmentation of molecules that occurs within the electrospray ionization (ESI) source of a mass spectrometer, primarily because of the high-energy environment (high temperature and voltage) present during the ionization process. This phenomenon has significant implications for metabolomics, as it can lead to the misannotation of fragments as genuine metabolites. ISF has been associated with the overrepresentation of peaks in LC–MS data, and some research suggests that ISF artifacts convolute the dark metabolome—the vast number of unidentified spectral features in untargeted metabolomics experiments. Studies have identified cases where in-source fragments mimic common metabolites, making accurate biological interpretation challenging (15).

Certain molecular structures are particularly susceptible to ISF, particularly compounds containing fragile moieties such as hydroxyl groups, lactones, glycosyl groups, and ether bonds. This susceptibility can result in characteristic neutral losses, such as water, carbon monoxide, or deglycosylation, potentially hampering identification and quantification. While there has been debate within the metabolomics community regarding the exact extent to which ISF contributes to a database—with some studies suggesting over 70% of features stemming from ISF in chemical standards vs. a more cautious view for complex biological samples—the consensus is that recognizing and addressing these artifacts is crucial for scientific rigor and preventing false discoveries (16).

To tackle ISF, a multifaceted approach combining experimental and computational strategies is necessary. Experimental validation relies on observing consistent intensity ratios and appearance patterns of true fragments across multiple samples, whereas contaminating fragments tend to exhibit more independent behavior. Cross-platform validation (comparing results across different instrumental platforms and ionization methods) also helps identify platform-specific artifacts.

Computational approaches are increasingly vital for comprehensive ISF detection (17), and machine learning algorithms show promise for automated ISF recognition by analyzing fragmentation patterns and metabolite relationships within large spectral databases. Furthermore, IM-MS plays a key role because it enables the separation of LC coeluting precursor ions in the drift time dimension prior to fragmentation. This capability enables the filtering out of fragments that do not match the precursor ion’s drift time, effectively removing in-source fragments, as demonstrated for tryptophan (12). This process helps in obtaining cleaner, high-quality spectra, decreasing false positive annotations.

In terms of future directions, there is an active area of research dedicated to the development of softer ionization methods and source designs that minimize unwanted fragmentation while maintaining analytical sensitivity. The metabolomics community is also working toward establishing standardized protocols for ISF assessment and reporting, including quality metrics and validation procedures, to ensure better comparability and reliability across studies. Finally, the integration of multiple analytical dimensions, including accurate mass, retention time, ion mobility, and fragmentation patterns, provides a more robust framework for distinguishing true metabolites from ISF artifacts.

Restricted Coverage with Data-dependent Acquisition

DDA is a common mass spectrometry technique where the instrument selectively targets and fragments specific ions based on their abundance or a similar pre-defined parameter. In a typical DDA workflow, a full MS scan is performed, followed by MS2 analysis on a predetermined list of precursor ions selected from the full-scan spectrum. While DDA allows for the simultaneous acquisition of both quantitative (from MS1) and structural (from MS2) information, it suffers from several inherent limitations that lead to restricted metabolome coverage, particularly in complex biological samples.

One major drawback of conventional DDA is its inherent bias: it prioritizes the most abundant ions for fragmentation, often neglecting or undersampling metabolites present at lower concentrations. This means that biologically significant but low-abundance metabolites may never be selected for fragmentation, resulting in incomplete MS2 data and missed identifications. This preference for abundant ions leads to lower reproducibility and incomplete coverage in untargeted analyses. Furthermore, DDA strategies allocate a significant portion of the instrument’s acquisition time to MS2 spectra generation, which consequently decreases the signal intensity for MS1 features, resulting in fewer points per LC peak as well as reduced sensitivity and quantitative accuracy.

In complex biological samples, the thousands of “features” detected by LC–MS often include a large proportion of non-biological signals or redundancies, such as contaminants from sample handling, artifacts from informatic errors or instrument noise, and degenerate signals of the same metabolite (adducts, oligomers, and naturally occurring isotopes) (18). By typically focusing on only a subset of the total ions, DDA can spend valuable acquisition time fragmenting these non-biological or redundant features, rather than unique biological metabolites. Even with the fastest mass spectrometers, it is impractical to individually target every single feature for MS/MS analysis using narrow isolation windows in a single chromatographic run. As a result, MS/MS spectra are often not collected for all metabolites present.

In contrast, DIA aims to capture all fragment ions systematically, providing an unbiased detection and quantification of every detectable analyte in a sample. DIA can generate comprehensive MS2 data for all precursor ions, potentially offering 100% MS/MS coverage in a single run, though often at the expense of generating complex and potentially chimeric spectra.

To improve the coverage and efficiency of MS/MS data acquisition, particularly for unique biological signals, several strategies have been developed to enhance DDA or move beyond its limitations:

• Intelligent exclusion lists can be used in iterative DDA runs to prevent re-fragmentation of already identified or highly abundant ions, allowing less intense metabolites to be targeted in subsequent runs. This can increase MS/MS coverage by approximately 25% over multiple runs.
• On-the-fly annotation strategies can identify and filter non-biological and redundant features in real-time during data acquisition. Simple data-processing techniques such as blank feature filtering and accurate-mass calculations can effectively remove thousands of contaminant and degenerate signals, decreasing the MS/MS burden by an order of magnitude. This allows for selective targeting of unique biological metabolites for MS/MS analysis.
• Pooled samples can be analyzed upfront to create a comprehensive list of potential features, ensuring that all unique biological signals are represented for annotation and subsequent targeted MS/MS analysis.
• The use of ion mobility to decrease reliance on quadrupole isolation can substantially increase acquisition speed without compromising sensitivity. This makes it possible to capture a significantly greater share of features, providing more extensive coverage.

These advancements move toward acquiring highly selective and comprehensive MS/MS data, focusing on biologically relevant molecules and decreasing the time and resources spent on non-informative signals (18).

Increasing Throughput for Large-Scale Studies

The ability to perform metabolomics on a large scale is increasingly crucial for advancing precision medicine, population health research, and biomarker discovery. Projects such as the UK Biobank’s initiative, involving over 280,000 participants, highlight both the immense potential and the significant challenges of scaling metabolomics to population-level studies. Large-scale cohort studies, high-temporal-resolution sampling, and single-cell analyses are now capable of generating thousands of ion features per sample, at throughputs 10 to 100 times higher than traditional workflows. However, the analytical and computational demands of processing such high volumes of data present considerable throughput limitations for conventional metabolomics approaches. Manual processing of thousands of features is simply impractical for large cohorts. To meet these demands, significant advancements in both analytical platforms and computational workflows are being implemented.

A demonstrated workflow for large-scale studies involves evaluating a reference sample (created by pooling aliquots from the cohort) to capture the chemical complexity of the biological matrix. This reference sample is processed with conventional software, and non-unique features are filtered out. The remaining features form a comprehensive set of biologically relevant reference chemicals that can then be efficiently extracted from the entire cohort’s raw data, based on m/z values and retention times. This strategy substantially decreases the computational burden, allowing informatics tools typically applied to targeted studies to be leveraged for profiling identified and unidentified features rapidly. This workflow has been successfully applied to analyze over 10,000 human plasma samples (19).

Quality control and standardization are paramount for generating reliable and reproducible results in large-scale metabolomics. This includes standardized preanalytical procedures for sample collection, processing, and storage. Implementing robust batch effect management techniques, such as random forest-based approaches, is essential for correcting technical variation between sample batches, particularly in longitudinal studies where full randomization may not be feasible. Quality control samples are regularly run to check instrument performance and correct for signal drift. International standards and collaborative frameworks such as COMETS provide crucial foundations for future large-scale initiatives (20). The ultimate goal is to enable metabolomics to meaningfully contribute to precision medicine and population health research by detecting subtle yet clinically significant associations that can inform disease prevention and treatment strategies.

Conclusion

The dark multi-ome includes the vast, uncharacterized molecular landscape within biological samples and presents a formidable frontier in metabolomics and related fields. Traditionally, inherent challenges such as isomeric ambiguity, the detection of low-abundance metabolites, the complexity of chimeric MS/MS spectra, the interference from in-source fragmentation artifacts, and the limitations of data-dependent acquisition have hampered comprehensive molecular characterization. However, the integration of cutting-edge mass spectrometry with IMS, specifically HRIM technology, offers transformative solutions.

These advanced technologies provide an orthogonal dimension of separation, enhancing the resolution of isomers and coeluting species, improving sensitivity for low-abundance compounds by amplifying signals, and simplifying complex chimeric spectra through deconvolution and fragment filtering. User-friendly computational tools are becoming increasingly available, and, alongside novel data processing workflows designed for population-scale studies, are critical for managing the immense datasets generated and extracting meaningful biological insights. The emphasis on standardization, quality control, and the development of robust bioinformatics frameworks further strengthens the reliability and reproducibility of large-scale omics investigations. While the dark multi-ome still holds many secrets, the synergistic development of advanced instrumentation and computational methods is rapidly illuminating this space, paving the way for deeper understanding of biological systems and significant advancements in precision medicine and biomarker discovery.

References

(1) Sindelar, M.; Patti, G. J. Chemical Discovery in the Era of Metabolomics. J. Am. Chem. Soc. 2020, 142 (20), 9097–9105. DOI: 10.1021/jacs.9b13198

(2) Blaženović, I.; Kind, T.; Ji, J.; Fiehn, O. Software Tools and Approaches for Compound Identification of LC-MS/MS Data in Metabolomics. Metabolites 2018, 8 (2), 31. DOI: 10.3390/metabo8020031

(3) Zhou, Z.; Luo, M.; Chen, X.; et al. Ion Mobility Collision Cross-section Atlas for Known and Unknown Metabolite Annotation in Untargeted Metabolomics. Nat. Commun. 2020, 11 (1), 4334. DOI: 10.1038/s41467-020-18171-8

(4) Koomen, D. C.; May, J. C.; McLean, J. A. Insights and Prospects for Ion Mobility-Mass Spectrometry in Clinical Chemistry. Expert Rev. Proteomics 2022, 19 (1), 17–31. DOI: 10.1080/14789450.2022.2026218

(5) Dodds, J. N.; Ford, L. C.; Ryan, J. P.; et al. Evaluating Ion Mobility Data Acquisition, Calibration, and Processing for Small Molecules: A Cross-Platform Assessment of Drift Tube and Traveling Wave Methodologies. J. Am. Soc. Mass Spectrom. 2025, 36 (7), 1456–1466. DOI: 10.1021/jasms.5c00056

(6) Deng, L.; Ibrahim, Y. M.; Baker, E. S.; et al. Ion Mobility Separations of Isomers based upon Long Path Length Structures for Lossless Ion Manipulations Combined with Mass Spectrometry. ChemistrySelect 2016, 1 (10), 2396–2399. DOI: 10.1002/slct.201600460

(7) Opialla, T.; Kempa, S.; Pietzke, M. Towards a More Reliable Identification of Isomeric Metabolites Using Pattern Guided Retention Validation. Metabolites 2020, 10 (11), 457. DOI: 10.3390/metabo10110457

(8) Onuh, J. O.; Qiu, H. Metabolic Profiling and Metabolites Fingerprints in Human Hypertension: Discovery and Potential. Metabolites 2021, 11 (10), 687. DOI: 10.3390/metabo11100687

(9) Lai, Y.; Koelmel, J. P.; Walker, D. I.; et al. High-Resolution Mass Spectrometry for Human Exposomics: Expanding Chemical Space Coverage. Environ. Sci. Technol. 2024, 58 (29), 12784–12822. DOI: 10.1021/acs.est.4c01156

(10) Qu, F.; Zhang, M.; Weinstock-Guttman, B.; et al. An Ultra-Sensitive and High-Throughput Trapping-Micro-LC-MS Method for Quantification of Circulating Vitamin D Metabolites and Application in Multiple Sclerosis Patients. Sci. Rep. 2024, 14 (1), 5545. DOI: 10.1038/s41598-024-55939-0

(11) Souza, A. L.; Patti, G. J. A Protocol for Untargeted Metabolomic Analysis: From Sample Preparation to Data Processing. Methods Mol. Biol. 2021, 2276, 357–382. DOI: 10.1007/978-1-0716-1266-8_27

(12) Pezzatti, J.; González-Ruiz, V.; Boccard, J.; Guillarme, D.; Rudaz, S. Evaluation of Different Tandem MS Acquisition Modes to Support Metabolite Annotation in Human Plasma Using Ultra High-Performance Liquid Chromatography High-Resolution Mass Spectrometry for Untargeted Metabolomics. Metabolites 2020, 10 (11), 464. DOI: 10.3390/metabo10110464

(13) Stancliffe, E.; Schwaiger-Haber, M.; Sindelar, M.; Patti, G. J. DecoID Improves Identification Rates in Metabolomics Through Database-Assisted MS/MS Deconvolution. Nat. Methods 2021, 18 (7), 779–787. DOI: 10.1038/s41592-021-01195-3

(14) de Jonge, N. F.; Hecht, H.; Strobel, M.; et al. Reproducible MS/MS Library Cleaning Pipeline in Matchms. J. Cheminform. 2024, 16 (1), 88. DOI: 10.1186/s13321-024-00878-1

(15) El Abiead, Y.; Rutz, A.; Zuffa, S.; et al. (2025). “Discovery of Metabolites Prevails Amid In-source Fragmentation. Nat. Metab. 2025, 7 (3), 435–437. DOI: 10.1038/s42255-025-01239-4

(16) Uritboonthai, W.; Hoang, L.; Aisporna, A.; Giera, M.; Siuzdak, G. The Dark Metabolome/Lipidome and In-Source Fragmentation. Anal. Sci. Adv. 2025, 6 (1), e70012. DOI: 10.1002/ansa.70012

(17) Guo, J.; Shen, S.; Xing, S.; Yu, H.; Huan, T. ISFrag: De Novo Recognition of In-Source Fragments for Liquid Chromatography–Mass Spectrometry Data. Anal. Chem. 2021, 93 (29), 10243–10250. DOI: 10.1021/acs.analchem.1c01644

(18) Cho, K.; Schwaiger-Haber, M.; Naser, F. J.; et al. Targeting Unique Biological Signals on the Fly to Improve MS/MS Coverage and Identification Efficiency in Metabolomics. Anal. Chim. Acta. 2021, 1149, 338210. DOI: 10.1016/j.aca.2021.338210

(19) Stancliffe, E.; Schwaiger-Haber, M.; Sindelar, M.; et al. An Untargeted Metabolomics Workflow that Scales to Thousands of Samples for Population-Based Studies. Anal. Chem. 2022, 94 (50), 17370–17378. DOI: 10.1021/acs.analchem.2c01270

(20) Yu, B.; Zanetti, K. A.; Temprosa, M.; et al. The Consortium of Metabolomics Studies (COMETS): Metabolomics in 47 Prospective Cohort Studies. Am. J. Epidemiol. 2019, 188 (6), 991–1012. DOI: 10.1093/aje/kwz028