IDSL_MINT: a deep learning framework to predict molecular fingerprints from mass spectra.

The majority of tandem mass spectrometry (MS/MS) spectra in untargeted metabolomics and exposomics studies lack any annotation. Our deep learning framework, Integrated Data Science Laboratory for Metabolomics and Exposomics-Mass INTerpreter (IDSL_MINT) can translate MS/MS spectra into molecular fingerprint descriptors. IDSL_MINT allows users to leverage the power of the transformer model for mass spectrometry data, similar to the large language models. Models are trained on user-provided reference MS/MS libraries via any customizable molecular fingerprint descriptors. IDSL_MINT was benchmarked using the LipidMaps database and improved the annotation rate of a test study for MS/MS spectra that were not originally annotated using existing mass spectral libraries. IDSL_MINT may improve the overall annotation rates in untargeted metabolomics and exposomics studies. The IDSL_MINT framework and tutorials are available in the GitHub repository at https://github.com/idslme/IDSL_MINT .Scientific contribution statement.Structural annotation of MS/MS spectra from untargeted metabolomics and exposomics datasets is a major bottleneck in gaining new biological insights. Machine learning models to convert spectra into molecular fingerprints can help in the annotation process. Here, we present IDSL_MINT, a new, easy-to-use and customizable deep-learning framework to train and utilize new models to predict molecular fingerprints from spectra for the compound annotation workflows.

IDSL.CSA: Composite Spectra Analysis for Chemical Annotation of Untargeted Metabolomics Datasets.

Poor chemical annotation of high-resolution mass spectrometry data limits applications of untargeted metabolomics datasets. Our new software, the Integrated Data Science Laboratory for Metabolomics and Exposomics─Composite Spectra Analysis (IDSL.CSA) R package, generates composite mass spectra libraries from MS1-only data, enabling the chemical annotation of high-resolution mass spectrometry coupled with liquid chromatography peaks regardless of the availability of MS2 fragmentation spectra. We demonstrate comparable annotation rates for commonly detected endogenous metabolites in human blood samples using IDSL.CSA libraries versus MS/MS libraries in validation tests. IDSL.CSA can create and search composite spectra libraries from any untargeted metabolomics dataset generated using high-resolution mass spectrometry coupled to liquid or gas chromatography instruments. The cross-applicability of these libraries across independent studies may provide access to new biological insights that may be missed due to the lack of MS2 fragmentation data. The IDSL.CSA package is available in the R-CRAN repository at https://cran.r-project.org/package=IDSL.CSA. Detailed documentation and tutorials are provided at https://github.com/idslme/IDSL.CSA.

IDSL.CSA: Composite Spectra Analysis for Chemical Annotation of Untargeted Metabolomics Datasets.

Poor chemical annotation of high-resolution mass spectrometry data limit applications of untargeted metabolomics datasets. Our new software, the Integrated Data Science Laboratory for Metabolomics and Exposomics - Composite Spectra Analysis (IDSL.CSA) R package, generates composite mass spectra libraries from MS1-only data, enabling the chemical annotation of LC/HRMS peaks regardless of the availability of MS2 fragmentation spectra. We demonstrate comparable annotation rates for commonly detected endogenous metabolites in human blood samples using IDSL.CSA libraries versus MS/MS libraries in validation tests. IDSL.CSA can create and search composite spectra libraries from any untargeted metabolomics dataset generated using high-resolution mass spectrometry coupled to liquid or gas chromatography instruments. The cross-applicability of these libraries across independent studies may provide access to new biological insights that may be missed due to the lack of MS2 fragmentation data. The IDSL.CSA package is available in the R CRAN repository at https://cran.r-project.org/package=IDSL.CSA . Detailed documentation and tutorials are provided at https://github.com/idslme/IDSL.CSA .

Bioaccumulation of perfluoroalkyl substances in the Lake Erie food web.

The bioaccumulation and biomagnification of perfluoroalkyl substances (PFAS) in the Lake Erie food web was investigated by analyzing surface water and biological samples including 10 taxa of fish species, 2 taxa of benthos and zooplankton. The carbon (δ13C) and nitrogen (δ15N) isotopic composition and fatty acids profiles of biological samples were used to evaluate the food web structure and assess the biomagnification of PFAS. Perfluorooctane sulfonate (PFOS) dominated the total PFAS (ΣPFAS) concentration (50-90% of ΣPFAS concentration), followed by C9-C11 perfluorinated carboxylic acids (PFCAs). The highest PFOS concentrations (79 ± 4.8 ng/g, wet weight (wwt)) and ΣPFAS (88 ± 5.2 ng/g, wwt) were detected in yellow perch (Perca flavescens). The C8-C14 PFAS biomagnification factors (BMFs) between apex piscivorous fish and prey fish were found to be generally greater than 1, indicative of PFAS biomagnification, while biodilution (BMF<1) was observed between planktivorous fish and zooplankton. Trophic magnification factors (TMFs) of C8-C14 PFCA were not correlated with perfluoroalkyl chain length. The C4-C9 PFAS were detected in the surface water of Lake Erie, and PFBA was found to have the highest concentrations (2.1-2.8 ng/L) among all PFAS detected. The log of bioaccumulation factor (BAF) was found to generally increase with increasing log Kow for C6, 8, and 9 PFAS in all selected species from three tropic levels.

IDSL.UFA Assigns High-Confidence Molecular Formula Annotations for Untargeted LC/HRMS Data Sets in Metabolomics and Exposomics.

Untargeted liquid chromatography/high-resolution mass spectrometry (LC/HRMS) assays in metabolomics and exposomics aim to characterize the small molecule chemical space in a biospecimen. To gain maximum biological insights from these data sets, LC/HRMS peaks should be annotated with chemical and functional information including molecular formula, structure, chemical class, and metabolic pathways. Among these, molecular formulas may be assigned to LC/HRMS peaks through matching theoretical and observed isotopic profiles (MS1) of the underlying ionized compound. For this, we have developed the Integrated Data Science Laboratory for Metabolomics and Exposomics-United Formula Annotation (IDSL.UFA) R package. In the untargeted metabolomics validation tests, IDSL.UFA assigned 54.31-85.51% molecular formula for true positive annotations as the top hit and 90.58-100% within the top five hits. Molecular formula annotations were also supported by tandem mass spectrometry data. We have implemented new strategies to (1) generate formula sources and their theoretical isotopic profiles, (2) optimize the formula hits ranking for the individual and aligned peak lists, and (3) scale IDSL.UFA-based workflows for studies with larger sample sizes. Annotating the raw data for a publicly available pregnancy metabolome study using IDSL.UFA highlighted hundreds of new pregnancy-related compounds and also suggested the presence of chlorinated perfluorotriether alcohols (Cl-PFTrEAs) in human specimens. IDSL.UFA is useful for human metabolomics and exposomics studies where we need to minimize the loss of biological insights in untargeted LC/HRMS data sets. The IDSL.UFA package is available in the R CRAN repository https://cran.r-project.org/package=IDSL.UFA. Detailed documentation and tutorials are also provided at www.ufa.idsl.me.

Bioaccumulation of polyfluoroalkyl substances in the Lake Huron aquatic food web.

Polyfluoroalkyl substances (PFAS) are a group of fluorinated organic chemicals that have been produced for industrial and commercial application since the 1950s. PFAS are highly persistent and ubiquitous in water, sediment, and biota. Toxic effects of PFAS on humans and the ecosystem have increased scientific and public concern. To better understand the distribution of PFAS in the Laurentian Great Lakes, carbon (12C and 13C) and nitrogen (14N and 15N) stable isotope enrichment, fatty acid profiles, and PFAS were measured in the Lake Huron (LH) aquatic food web. The trophic level of the organisms was estimated using δ15N and found to be a determinant of PFAS biomagnification. The δ13C and fatty acid profiles were used to assess the carbon/energy flow pathway and predator-prey relationships, respectively. The δ13C, δ15N, and fatty acids were used to elucidate the trophodynamics and understand the PFAS trophic transfer in the LH aquatic food web. Perfluorooctanesulfonic acid (PFOS) was the dominant PFAS observed, followed by C9 - C11 perfluorinated carboxylic acids (PFCA). The highest PFOS concentrations (45 ± 11 ng/g, wet weight (wwt)) were detected in lake trout (Salvelinus namaycush), while the highest total PFCA concentrations (sum of C4 - C16 PFCAs) were detected in deepwater sculpin (Myoxocephalus thompsonii). With the exception of perfluorooctanoic acid (PFOA), C8-C14 PFAS biomagnification factors (BMFs) were found to be generally greater than 1, suggesting PFAS biomagnification from prey to predator. Trophic magnification factors (TMFs) of C8-C14 PFCA were found to be independent of compound hydrophobicity.

Data Processing Thresholds for Abundance and Sparsity and Missed Biological Insights in an Untargeted Chemical Analysis of Blood Specimens for Exposomics.

Background: An untargeted chemical analysis of bio-fluids provides semi-quantitative data for thousands of chemicals for expanding our understanding about relationships among metabolic pathways, diseases, phenotypes and exposures. During the processing of mass spectral and chromatography data, various signal thresholds are used to control the number of peaks in the final data matrix that is used for statistical analyses. However, commonly used stringent thresholds generate constrained data matrices which may under-represent the detected chemical space, leading to missed biological insights in the exposome research. Methods: We have re-analyzed a liquid chromatography high resolution mass spectrometry data set for a publicly available epidemiology study (n = 499) of human cord blood samples using the MS-DIAL software with minimally possible thresholds during the data processing steps. Peak list for individual files and the data matrix after alignment and gap-filling steps were summarized for different peak height and detection frequency thresholds. Correlations between birth weight and LC/MS peaks in the newly generated data matrix were computed using the spearman correlation coefficient. Results: MS-DIAL software detected on average 23,156 peaks for individual LC/MS file and 63,393 peaks in the aligned peak table. A combination of peak height and detection frequency thresholds that was used in the original publication at the individual file and the peak alignment levels can reject 90% peaks from the untargeted chemical analysis dataset that was generated by MS-DIAL. Correlation analysis for birth weight data suggested that up to 80% of the significantly associated peaks were rejected by the data processing thresholds that were used in the original publication. The re-analysis with minimum possible thresholds recovered metabolic insights about C19 steroids and hydroxy-acyl-carnitines and their relationships with birth weight. Conclusions: Data processing thresholds for peak height and detection frequencies at individual data file and at the alignment level should be used at minimal possible level or completely avoided for mining untargeted chemical analysis data in the exposome research for discovering new biomarkers and mechanisms.

Nontargeted Discovery of Novel Contaminants in the Great Lakes Region: A Comparison of Fish Fillets and Fish Consumers.

Sport fish fillets and human sera (fish consumers) were collected in the Lake Superior and Lake Michigan basin and screened for novel contaminants using the isotopic profile deconvoluted chromatogram (IPDC) algorithm. The IPDC algorithm was extended beyond traditional Cl/Br filters to detect additional potential bioaccumulative and toxic (PBT) such as perfluoroalkyl substances (PFAS). The IPDC algorithm screened for approximately 13.5 million theoretical molecular formulas. Additional algorithm modules were developed to detect data independent MS/MS fragmentation products and a retention time index calculator using a series of 13C-labeled perfluoroalkyl carboxylic acids (13C-PFCAs). Ten potential compound classes were isolated including six untargeted PFAS, six homologue groups of polyfluorinated carboxylic acids, polyfluorinated telomer alcohols (PoFTOHs), two hydroxylated polychlorobiphenyls, pesticides, herbicides, antifungals, pharmaceuticals, artificial sweeteners, and personal care products with minimal postprocessing efforts. The algorithm isolated 48 ubiquitous PoFTOHs in both fish fillet and serum of fish consumers suggesting a region wide distribution of this class of compounds. The 3, 4, and 7 fluorine substituted PoFTOH were the most abundant congeners in both biological matrices.

Breakdown Products from Perfluorinated Alkyl Substances (PFAS) Degradation in a Plasma-Based Water Treatment Process.

Byproducts produced when treating perfluorooctanoic acid (PFOA) and perfluorooctanesulfonate (PFOS) in water using a plasma treatment process intentionally operated to treat these compounds slowly to allow for byproduct accumulation were quantified. Several linear chain perfluoroalkyl carboxylic acids (PFCAs) (C4 to C7) were identified as byproducts of both PFOA and PFOS treatment. PFOA, perfluorohexanesulfonate (PFHxS), and perfluorobutanesulfonate (PFBS) were also found to be byproducts from PFOS degradation. Significant concentrations of fluoride ions, inorganic carbon, and smaller organic acids (trifluoroacetic acid, acetic acid, and formic acid) were also identified. In addition to PFCAs, PFHxS, and PFBS, trace amounts of 43 PFOA-related and 35 PFOS-related byproducts were also identified using a screening and search-based algorithm. Minor concentrations of gas-phase byproducts were also identified (<2.5% of the F originally associated with the parent molecules) some of which are reported for the first time in perfluoroalkyl substance degradation experiments including cyclic perfluoroalkanes (C4F8, C5F10, C6F12, C7F14, and C8F16). The short chain PFCAs detected suggest the occurrence of a stepwise reduction of the parent perfluoroalkyl substances (PFAS) molecule, followed by oxidation of intermediates, perfluoroalkyl radicals, and perfluoro alcohols/ketones. Using a fluorine mass balance, 77% of the fluorine associated with the parent PFOA and 58% of the fluorine associated with the parent PFOS were identified. The bulk of the remaining fluorine was determined to be sorbed to reactor walls and tubing using sorption experiments in which plasma was not generated.