Scientists have launched an open mass-spectrometry library designed to help researchers detect hundreds of chemical exposures that routinely disappear into the blind spots of modern metabolomics. Called ExpoLib, the resource contains representative fragmentation spectra for more than 170 naturally occurring and human-made toxicants, together with selected products formed when those compounds are transformed inside the body. The library is intended for exposomics—the study of the complete collection of chemicals to which people are exposed—and could make it easier to identify pollutants, food toxins, pesticide residues, drug compounds, consumer-product ingredients and biologically active metabolites in blood and other samples. The work, published in Metabolomics, addresses a central problem in untargeted chemical analysis: instruments can detect thousands of molecular signals, but many cannot be assigned a reliable identity. Without a matching reference spectrum, a signal may remain an anonymous feature even when it belongs to a compound with important toxicological effects. ExpoLib’s developers say the new collection is designed to expand the chemical space available for high-confidence annotation, particularly for compounds that are poorly represented in widely used public databases.
The challenge begins with the way liquid chromatography–tandem mass spectrometry, or LC–MS/MS, identifies molecules. In a typical experiment, compounds are first separated by liquid chromatography and then converted into charged ions by electrospray ionization. The instrument measures the mass-to-charge ratio of an intact precursor ion before colliding it with gas molecules and breaking it into smaller product ions. The resulting pattern, known as an MS/MS spectrum, acts like a molecular fingerprint. Researchers compare that pattern with spectra generated from known reference standards or stored in spectral libraries. A strong match can provide compelling evidence for a compound’s identity, but a missing or incomplete library entry can produce a false assignment—or no assignment at all. The problem is especially severe in exposomics because the relevant chemicals are extraordinarily diverse. They include industrial compounds, endocrine-disrupting chemicals, contaminants, pesticides, medicines, food additives, microbial toxins and metabolites generated by human or microbial enzymes. Although chemical databases contain many millions of structures, only a small fraction have experimentally measured fragmentation spectra under conditions useful for biological analysis.
ExpoLib was assembled to target this gap with a deliberately broad selection of compounds. Its collection includes natural toxins made by bacteria, fungi and plants, alongside anthropogenic chemicals such as bisphenols, phthalates, per- and polyfluoroalkyl substances, pesticides, pharmaceuticals and ingredients used in consumer-care products. The researchers also included toxicologically important transformation products for which commercial standards or publicly available MS/MS spectra are difficult or impossible to obtain. Examples include a colibactin–DNA adduct associated with bacterial genotoxicity, the food-poisoning toxin cereulide, and deoxynivalenol-3-glucuronide, a metabolite formed during the body’s processing of the mycotoxin deoxynivalenol. Such compounds are scientifically important precisely because they may reveal exposure or biological damage, yet they are likely to be overlooked by databases focused on more familiar metabolites. By incorporating both parent chemicals and biotransformation products, the library is intended to connect environmental exposure with the molecular traces left in human samples.
To generate the resource, the team analysed reference standards on an Agilent Infinity II 1290 liquid-chromatography system coupled to a SCIEX ZenoTOF 7600 high-resolution mass spectrometer. The separation used a reversed-phase HSS T3 column and an in-house method previously optimized for more than 200 compounds. The instrument operated in data-dependent acquisition mode, selecting precursor ions for fragmentation and recording spectra in both positive and negative electrospray ionization. Rather than relying on a single collision-energy setting, the researchers collected data from 10 to 70 volts in 5-volt steps, creating 13 single-energy files for each polarity. They also tested four collision-energy-spread settings: 30 ± 10, 30 ± 20, 40 ± 10 and 40 ± 20 volts. Collision energy controls how forcefully precursor ions are struck and therefore which chemical bonds break. Low energies may leave molecules insufficiently fragmented, while high energies can produce extensive breakdown patterns. Recording a range of energies allows users to select spectra that best resemble those generated by their own instruments and methods.
The team then performed extensive quality checks before releasing the final libraries. Each compound was evaluated for retention time, signal quality and mass accuracy, and spectra were reviewed to determine whether the expected molecular ion was present and whether interference or chimeric signals could compromise identification. The resulting data capture not only the familiar protonated and deprotonated forms of molecules—[M+H]+ and [M−H]−—but also alternative adducts. In positive mode, these included sodium and ammonium adducts as well as ions formed through the loss of water; in negative mode, chloride adducts were represented. Adducts arise when molecules associate with ions or lose small chemical groups during ionization, and they can determine whether a compound is visible or correctly recognized. More than one-third of the compounds showed useful ionization in both positive and negative modes, supporting the researchers’ argument that exposomics libraries should not be built around only the most efficient polarity for each molecule. A compound that produces a weak signal in one mode may generate a highly informative fragmentation pattern in the other.
Benchmarking showed that the choice of collision energy influenced the library’s performance, although the differences were generally modest across the most useful settings. In positive mode, collision energies of 35 and 30 volts produced the highest reported accuracy values, 78 and 77 percent, respectively. In negative mode, performance peaked at 40 volts and 40 ± 10 volts, with reported accuracies of 77 and 75 percent. Positive-mode precision reached 82 percent overall, compared with 67 percent in negative mode, reflecting the broader chemical coverage and stronger ratio of correct to incorrect matches in the positive library. Sensitivity—the ability to recover compounds that are actually present—reached 80 percent for positive-mode spectra at 30 ± 10 volts and 85 percent for negative-mode spectra at 25 volts. The researchers found that collision-energy-spread spectra could be beneficial, producing similar or better results than some single-energy measurements. However, mixed spectra can also be harder to match against databases dominated by single-energy data, and dividing acquisition time among several energies may increase noise. Spectra collected at 10 and 15 volts performed poorly, with all measured indicators below 70 percent, so those settings were excluded from the downloadable libraries.
The most revealing test used human serum samples deliberately spiked with known xenobiotics. First, the spectra were searched against selected open-source databases without ExpoLib. The process was then repeated after ExpoLib had been added. Performance improved in both ionization modes, but the benefit was especially pronounced in negative mode, where the new library substantially improved precision and reduced the false-discovery rate. Of 29 compounds that initially had no available MS/MS spectra in the comparison databases, most were converted from unidentifiable signals into correct matches after ExpoLib was included. The library also corrected several misleading results: 19 initial false-positive assignments were reduced to four after the expanded database enabled more appropriate spectral matches. In positive mode, accuracy and precision rose considerably, while negative mode showed broad improvements across the evaluated metrics. The serum experiment does not prove that the library will perform identically for every biological matrix or instrument, and the authors caution that the benchmark represents only a small portion of chemical space. Still, the results demonstrate how missing reference spectra can distort non-targeted analysis and how adding carefully measured entries can rescue signals that would otherwise remain invisible.
The researchers have released the raw data and library files through the Global Exposomics and Biomonitoring Lab repository on Zenodo. Users can download files in MSP format for MS-DIAL, JSON format recommended for MZmine, and SDF format suitable for Library View and SCIEX OS workflows. Separate positive- and negative-ion versions are provided, including restricted files containing only the dominant protonated or deprotonated adducts and broader versions containing additional adduct types. The release also includes compound metadata, retention times, collision-energy information, quality-control documentation, MZmine presets and batch files, a database template and an R script for producing library-overview files. This emphasis on reproducibility is central to the project. The authors describe the complete workflow so that other laboratories can generate their own in-house libraries from the same raw measurements, adapt the processing parameters and extend the collection with additional chemicals. They argue that spectral-library development should routinely include transparent quality assurance, standardized reporting and benchmarking against independent or in-house samples. Such measures will be increasingly important as exposomics moves from exploratory surveys toward clinical, environmental and public-health applications.
ExpoLib does not eliminate the need for chemical confirmation. The strongest identification still comes from matching an unknown sample’s accurate mass, retention time and fragmentation spectrum with those of an authentic reference standard analysed under comparable conditions. Nor can any single library represent the vast and changing universe of synthetic chemicals, natural products and human metabolites. The resource nevertheless offers a practical way to improve the first and often most difficult stage of discovery: recognizing which molecular signals deserve attention. Better annotation could help researchers trace dietary toxins, monitor persistent pollutants, identify microbial genotoxins and examine how exposure-related compounds are altered by human metabolism. It may also support retention-time prediction and in-silico fragmentation modelling, extending the library’s usefulness beyond direct spectrum matching. As untargeted metabolomics increasingly encounters chemicals that were never part of conventional metabolic studies, the ability to distinguish meaningful toxicants from anonymous background signals could become a decisive factor in understanding how the environment affects health.
Subject of Research: An open MS/MS spectral library for identifying anthropogenic and natural toxicants, xenobiotics, and their biotransformation products in exposomics and metabolomics research.
Subject of Research: Biology
Article Title: ExpoLib: a framework for an MS/MS exposome library of anthropogenic and natural toxicants and their biotransformation products
Article References: Hernandes, V. V., Ramos, M. A. A., Breinbauer, R., Fruhmann, P., Mikula, H., Ehling-Schulz, M., Zechner, E. L., Balskus, E. P., & Warth, B. (2026). ExpoLib: a framework for an MS/MS exposome library of anthropogenic and natural toxicants and their biotransformation products. Metabolomics, 22(5), Article 145. https://doi.org/10.1007/s11306-026-02481-x
Image Credits: AI Generated
DOI: 10.1007/s11306-026-02481-x
Keywords: exposomics, high-resolution mass spectrometry, MS/MS spectral library, toxicants, xenobiotics, biotransformation products, environmental contaminants, non-targeted screening, metabolomics
Cite this news
APA MLA Chicago
SCIENMAG. (August 28, 2026). ExpoLib Framework Builds MS/MS Exposome Library Tracking Toxicants and Biotransformation Products. https://scienmag.com/expolib-framework-builds-ms-ms-exposome-library-tracking-toxicants-and-biotransformation-products/
SCIENMAG. “ExpoLib Framework Builds MS/MS Exposome Library Tracking Toxicants and Biotransformation Products.” Scienmag, 28 August 2026, https://scienmag.com/expolib-framework-builds-ms-ms-exposome-library-tracking-toxicants-and-biotransformation-products/. Accessed 28 August 2026.
SCIENMAG. “ExpoLib Framework Builds MS/MS Exposome Library Tracking Toxicants and Biotransformation Products.” Scienmag. August 28, 2026. https://scienmag.com/expolib-framework-builds-ms-ms-exposome-library-tracking-toxicants-and-biotransformation-products/
Copy citation Download RIS
Tags: biologically active metabolite profilingbiologically active metabolites detectionbiotransformation product analysischemical exposure detectionexpanding chemical space in metabolomicsExpoLibExpoLib mass spectrometry libraryexposomics research toolsfood contaminant detectionhigh-confidence chemical annotationidentification of food toxins and pesticide residuesLC–MS/MS molecule identificationLC–MS/MS spectral databasesmass spectrometry exposome librarymetabolomics of environmental pollutantspesticide residue analysispollutant and toxin identificationreference spectra for toxic chemicalstoxicant identification in metabolomicstoxicants and biotransformation productsuntargeted chemical analysis
