In its fourth year, the CASMI 2016 contest was organized to evaluate current chemical structure identification strategies for 19 natural products using high-resolution LC-MS and LC-MS/MS challenge datasets using automated methods with or without the combination of other tools. These natural products originate from plants, fungi, marine sponges, algae, or micro-algae. Every compound annotation workflow must start with determination of elemental compositions. Of these 19 challenges, one was excluded by the organizers after submission. For the remaining 18 challenges, three software programs were used. MS-FINDER version 1.62 was able to correctly identify 89% of the molecular formulas using an internal database that comprised of 13 metabolomics repositories with 45,181 formulas. SIRIUS correctly identified 61% compositions using PubChem formulas and Seven Golden Rules correctly identified 83% by using the Dictionary of Natural Products as a targeted database. Next, we performed structural dereplication for which we used the consensus formula from the three software programs. We submitted two solution sets for these challenges. In the first solution set, avaniya001, we only used the internal MS-FINDER functions for predicting and ranking structures, correctly identifying 53% of the structures as top-hit, 72% within the top-3 structures, and 78% within the top-10 hits. For our second set, avaniya002, we used both MS-FINDER predictions as well as MS/MS queries against the commercial NIST 14, METLIN, and the public MassBank of North America libraries. Here we correctly identified 78% of the structures as top-hit and 83% within the top-3 hits. Three challenge spectra remained unidentified in either of our submissions within the top-10 hits.

Unidentified peaks remain a major problem in untargeted metabolomics by LC-MS/MS. Confidence in peak annotations increases by combining MS/MS matching and retention time. We here show how retention times can be predicted from molecular structures. Two large, publicly available data sets were used for model training in machine learning: the Fiehn hydrophilic interaction liquid chromatography data set (HILIC) of 981 primary metabolites and biogenic amines,and the RIKEN plant specialized metabolome annotation (PlaSMA) database of 852 secondary metabolites that uses reversed-phase liquid chromatography (RPLC). Five different machine learning algorithms have been integrated into the Retip R package: the random forest, Bayesian-regularized neural network, XGBoost, light gradient-boosting machine (LightGBM), and Keras algorithms for building the retention time prediction models. A complete workflow for retention time prediction was developed in R. It can be freely downloaded from the GitHub repository (https://www.retip.app). Keras outperformed other machine learning algorithms in the test set with minimum overfitting, verified by small error differences between training, test, and validation sets. Keras yielded a mean absolute error of 0.78 min for HILIC and 0.57 min for RPLC. Retip is integrated into the mass spectrometry software tools MS-DIAL and MS-FINDER, allowing a complete compound annotation workflow. In a test application on mouse blood plasma samples, we found a 68% reduction in the number of candidate structures when searching all isomers in MS-FINDER compound identification software. Retention time prediction increases the identification rate in liquid chromatography and subsequently leads to an improved biological interpretation of metabolomics data.

Alzheimer's disease (AD) is a neurodegenerative disease that commonly causes dementia. Identifying biomarkers for the early detection of AD is an emerging need, as brain dysfunction begins two decades before the onset of clinical symptoms. To this end, we reanalyzed untargeted metabolomic mass spectrometry data from 905 patients enrolled in the AD Neuroimaging Initiative (ADNI) cohort using MS-DIAL, with 1,304,633 spectra of 39,108 unique biomolecules. Metabolic profiles of 93 hydrophilic metabolites were determined. Additionally, we integrated targeted lipidomic data (4873 samples from 1524 patients) to explore candidate biomarkers for predicting progressive mild cognitive impairment (pMCI) in patients diagnosed with AD within two years using the baseline metabolome. Patients with lower ergothioneine levels had a 12% higher rate of AD progression with the significance of P = 0.012 (Wald test). Furthermore, an increase in ganglioside (GM3) and decrease in plasmalogen lipids, many of which are associated with apolipoprotein E polymorphism, were confirmed in AD patients, and the higher levels of lysophosphatidylcholine (18:1) and GM3 d18:1/20:0 showed 19% and 17% higher rates of AD progression, respectively (Wald test: P = 3.9 × 10^-8 and 4.3 × 10^-7). Palmitoleamide, oleamide, diacylglycerols, and ether lipids were also identified as significantly altered metabolites at baseline in patients with pMCI. The integrated analysis of metabolites and genomics data showed that combining information on metabolites and genotypes enhances the predictive performance of AD progression, suggesting that metabolomics is essential to complement genomic data. In conclusion, the reanalysis of multiomics data provides new insights to detect early development of AD pathology and to partially understand metabolic changes in age-related onset of AD.

Compound identification from accurate mass MS/MS spectra is a bottleneck for untargeted metabolomics. In this study, we propose nine rules of hydrogen rearrangement (HR) during bond cleavages in low-energy collision-induced dissociation (CID). These rules are based on the classic even-electron rule and cover heteroatoms and multistage fragmentation. We evaluated our HR rules by the statistics of MassBank MS/MS spectra in addition to enthalpy calculations, yielding three levels of computational MS/MS annotation: "resolved" (regular HR behavior following HR rules), "semiresolved" (irregular HR behavior), and "formula-assigned" (lacking structure assignment). With this nomenclature, 78.4% of a total of 18506 MS/MS fragment ions in the MassBank database and 84.8% of a total of 36370 MS/MS fragment ions in the GNPS database were (semi-) resolved by predicted bond cleavages. We also introduce the MS-FINDER software for structure elucidation. Molecular formulas of precursor ions are determined from accurate mass, isotope ratio, and product ion information. All isomer structures of the predicted formula are retrieved from metabolome databases, and MS/MS fragmentations are predicted in silico. The structures are ranked by a combined weighting score considering bond dissociation energies, mass accuracies, fragment linkages, and, most importantly, nine HR rules. The program was validated by its ability to correctly calculate molecular formulas with 98.0% accuracy for 5063 MassBank MS/MS records and to yield the correct structural isomer with 82.1% accuracy within the top-3 candidates. In a test with 936 manually identified spectra from an untargeted HILIC-QTOF MS data set of human plasma, formulas were correctly predicted in 90.4% of the cases, and the correct isomer structure was retrieved at 80.4% probability within the top-3 candidates, including for compounds that were absent in mass spectral libraries. The MS-FINDER software is freely available at http://prime.psc.riken.jp/ .

Data-independent acquisition (DIA) in liquid chromatography (LC) coupled to tandem mass spectrometry (MS/MS) provides comprehensive untargeted acquisition of molecular data. We provide an open-source software pipeline, which we call MS-DIAL, for DIA-based identification and quantification of small molecules by mass spectral deconvolution. For a reversed-phase LC-MS/MS analysis of nine algal strains, MS-DIAL using an enriched LipidBlast library identified 1,023 lipid compounds, highlighting the chemotaxonomic relationships between the algal strains.

Tandem mass spectral library search (MS/MS) is the fastest way to correctly annotate MS/MS spectra from screening small molecules in fields such as environmental analysis, drug screening, lipid analysis, and metabolomics. The confidence in MS/MS-based annotation of chemical structures is impacted by instrumental settings and requirements, data acquisition modes including data-dependent and data-independent methods, library scoring algorithms, as well as post-curation steps. We critically discuss parameters that influence search results, such as mass accuracy, precursor ion isolation width, intensity thresholds, centroiding algorithms, and acquisition speed. A range of publicly and commercially available MS/MS databases such as NIST, MassBank, MoNA, LipidBlast, Wiley MSforID, and METLIN are surveyed. In addition, software tools including NIST MS Search, MS-DIAL, Mass Frontier, SmileMS, Mass++, and XCMS² to perform fast MS/MS search are discussed. MS/MS scoring algorithms and challenges during compound annotation are reviewed. Advanced methods such as the in silico generation of tandem mass spectra using quantum chemistry and machine learning methods are covered. Community efforts for curation and sharing of tandem mass spectra that will allow for faster distribution of scientific discoveries are discussed.

Novel metabolites distinct from canonical pathways can be identified through the integration of three cheminformatics tools: BinVestigate, which queries the BinBase gas chromatography-mass spectrometry (GC-MS) metabolome database to match unknowns with biological metadata across over 110,000 samples; MS-DIAL 2.0, a software tool for chromatographic deconvolution of high-resolution GC-MS or liquid chromatography-mass spectrometry (LC-MS); and MS-FINDER 2.0, a structure-elucidation program that uses a combination of 14 metabolome databases in addition to an enzyme promiscuity library. We showcase our workflow by annotating N-methyl-uridine monophosphate (UMP), lysomonogalactosyl-monopalmitin, N-methylalanine, and two propofol derivatives.

In mass spectrometry-based untargeted metabolomics, rarely more than 30% of the compounds are identified. Without the true identity of these molecules it is impossible to draw conclusions about the biological mechanisms, pathway relationships and provenance of compounds. The only way at present to address this discrepancy is to use in silico fragmentation software to identify unknown compounds by comparing and ranking theoretical MS/MS fragmentations from target structures to experimental tandem mass spectra (MS/MS). We compared the performance of four publicly available in silico fragmentation algorithms (MetFragCL, CFM-ID, MAGMa+ and MS-FINDER) that participated in the 2016 CASMI challenge. We found that optimizing the use of metadata, weighting factors and the manner of combining different tools eventually defined the ultimate outcomes of each method. We comprehensively analysed how outcomes of different tools could be combined and reached a final success rate of 93% for the training data, and 87% for the challenge data, using a combination of MAGMa+, CFM-ID and compound importance information along with MS/MS matching. Matching MS/MS spectra against the MS/MS libraries without using any in silico tool yielded 60% correct hits, showing that the use of in silico methods is still important.

BACKGROUND:The fourth round of the Critical Assessment of Small Molecule Identification (CASMI) Contest ( www.casmi-contest.org ) was held in 2016, with two new categories for automated methods. This article covers the 208 challenges in Categories 2 and 3, without and with metadata, from organization, participation, results and post-contest evaluation of CASMI 2016 through to perspectives for future contests and small molecule annotation/identification. RESULTS:The Input Output Kernel Regression (CSI:IOKR) machine learning approach performed best in "Category 2: Best Automatic Structural Identification-In Silico Fragmentation Only", won by Team Brouard with 41% challenge wins. The winner of "Category 3: Best Automatic Structural Identification-Full Information" was Team Kind (MS-FINDER), with 76% challenge wins. The best methods were able to achieve over 30% Top 1 ranks in Category 2, with all methods ranking the correct candidate in the Top 10 in around 50% of challenges. This success rate rose to 70% Top 1 ranks in Category 3, with candidates in the Top 10 in over 80% of the challenges. The machine learning and chemistry-based approaches are shown to perform in complementary ways. CONCLUSIONS:The improvement in (semi-)automated fragmentation methods for small molecule identification has been substantial. The achieved high rates of correct candidates in the Top 1 and Top 10, despite large candidate numbers, open up great possibilities for high-throughput annotation of untargeted analysis for "known unknowns". As more high quality training data becomes available, the improvements in machine learning methods will likely continue, but the alternative approaches still provide valuable complementary information. Improved integration of experimental context will also improve identification success further for "real life" annotations. The true "unknown unknowns" remain to be evaluated in future CASMI contests. Graphical abstract .

The concept of the exposome was introduced over 15 years ago to reflect the important role that the environment exerts on health and disease. While originally viewed as a call-to-arms to develop more comprehensive exposure assessment methods applicable at the individual level and throughout the life course, the scope of the exposome has now expanded to include the associated biological response. In order to explore these concepts, a workshop was hosted by the Gunma University Initiative for Advanced Research (GIAR, Japan) to discuss the scope of exposomics from an international and multidisciplinary perspective. This Global Perspective is a summary of the discussions with emphasis on (1) top-down, bottom-up, and functional approaches to exposomics, (2) the need for integration and standardization of LC- and GC-based high-resolution mass spectrometry methods for untargeted exposome analyses, (3) the design of an exposomics study, (4) the requirement for open science workflows including mass spectral libraries and public databases, (5) the necessity for large investments in mass spectrometry infrastructure in order to sequence the exposome, and (6) the role of the exposome in precision medicine and nutrition to create personalized environmental exposure profiles. Recommendations are made on key issues to encourage continued advancement and cooperation in exposomics.