The potential of untargeted metabolomics to answer important questions across the life sciences is hindered because of a paucity of computational tools that enable extraction of key biochemically relevant information. Available tools focus on using mass spectrometry fragmentation spectra to identify molecules whose behavior suggests they are relevant to the system under study. Unfortunately, fragmentation spectra cannot identify molecules in isolation but require authentic standards or databases of known fragmented molecules. Fragmentation spectra are, however, replete with information pertaining to the biochemical processes present, much of which is currently neglected. Here, we present an analytical workflow that exploits all fragmentation data from a given experiment to extract biochemically relevant features in an unsupervised manner. We demonstrate that an algorithm originally used for text mining, latent Dirichlet allocation, can be adapted to handle metabolomics datasets. Our approach extracts biochemically relevant molecular substructures ("Mass2Motifs") from spectra as sets of co-occurring molecular fragments and neutral losses. The analysis allows us to isolate molecular substructures, whose presence allows molecules to be grouped based on shared substructures regardless of classical spectral similarity. These substructures, in turn, support putative de novo structural annotation of molecules. Combining this spectral connectivity to orthogonal correlations (e.g., common abundance changes under system perturbation) significantly enhances our ability to provide mechanistic explanations for biological behavior.metabolomics | mass spectrometry | fragmentation | bioinformatics | topic modeling M ass spectrometry (MS)-based metabolomics aims to capture the entire small-molecule composition of biological systems. Analysis of MS metabolomics data are challenging as many molecules cannot be identified from their mass (e.g., isobaric molecules, and isomers) (1-3). Separation by liquid chromatography before MS (LC-MS) can add discriminatory information but does not solve the problem as isomers can exhibit similar chromatographic behavior, and chromatographic retention time is currently unpredictable.Fragmentation spectra have been used to partially overcome this problem (4-6). Most tools compare individual fragmentation spectra to reference spectra (5, 7) stored in public databases, for example, MassBank (8) or Human Metabolome Database (9), and are thus constrained by the limited number of reference spectra (10-12). Poor identification coverage can result in poor biochemical insight. We propose a method that analyzes all acquired fragmentation spectra to expose underlying biochemistry without relying on metabolite identification, inspired by machine-learning techniques developed initially for text processing (13).The paucity of techniques that share information across fragmentation spectra can be explained by the complexity of fragmentation data (14). One example, "Molecular Networking," clusters MS1 peaks by th...

Metabolomics has started to embrace computational approaches for chemical interpretation of large data sets. Yet, metabolite annotation remains a key challenge. Recently, molecular networking and MS2LDA emerged as molecular mining tools that find molecular families and substructures in mass spectrometry fragmentation data. Moreover, in silico annotation tools obtain and rank candidate molecules for fragmentation spectra. Ideally, all structural information obtained and inferred from these computational tools could be combined to increase the resulting chemical insight one can obtain from a data set. However, integration is currently hampered as each tool has its own output format and efficient matching of data across these tools is lacking. Here, we introduce MolNetEnhancer, a workflow that combines the outputs from molecular networking, MS2LDA, in silico annotation tools (such as Network Annotation Propagation or DEREPLICATOR), and the automated chemical classification through ClassyFire to provide a more comprehensive chemical overview of metabolomics data whilst at the same time illuminating structural details for each fragmentation spectrum. We present examples from four plant and bacterial case studies and show how MolNetEnhancer enables the chemical annotation, visualization, and discovery of the subtle substructural diversity within molecular families. We conclude that MolNetEnhancer is a useful tool that greatly assists the metabolomics researcher in deciphering the metabolome through combination of multiple independent in silico pipelines.

MotivationWe recently published MS2LDA, a method for the decomposition of sets of molecular fragment data derived from large metabolomics experiments. To make the method more widely available to the community, here we present ms2lda.org, a web application that allows users to upload their data, run MS2LDA analyses and explore the results through interactive visualizations.ResultsMs2lda.org takes tandem mass spectrometry data in many standard formats and allows the user to infer the sets of fragment and neutral loss features that co-occur together (Mass2Motifs). As an alternative workflow, the user can also decompose a data set onto predefined Mass2Motifs. This is accomplished through the web interface or programmatically from our web service.Availability and implementationThe website can be found at http://ms2lda.org, while the source code is available at https://github.com/sdrogers/ms2ldaviz under the MIT license.Supplementary information Supplementary data are available at Bioinformatics online.

Metabolomics has started to embrace computational approaches for chemical interpretation of large data sets. Yet, metabolite annotation remains a key challenge. Recently, molecular networking and MS2LDA emerged as molecular mining tools that find molecular families and substructures in mass spectrometry fragmentation data. Moreover, in silico annotation tools obtain and rank candidate molecules for fragmentation spectra. Ideally, all structural information obtained and inferred from these computational tools could be combined to increase the resulting chemical insight one can obtain from a data set. However, integration is currently hampered as each tool has its own output format and efficient matching of data across these tools is lacking. Here, we introduce MolNetEnhancer, a workflow that combines the outputs from molecular networking, MS2LDA, in silico annotation tools (such as Network Annotation Propagation or DEREPLICATOR) and the automated chemical classification through ClassyFire to provide a more comprehensive chemical overview of metabolomics data whilst at the same time illuminating structural details for each fragmentation spectrum. We present examples from four plant and bacterial case studies and show how MolNetEnhancer enables the chemical annotation, visualization, and discovery of the subtle substructural diversity within molecular families. We conclude that MolNetEnhancer is a useful tool that greatly assists the metabolomics researcher in deciphering the metabolome through combination of multiple independent in silico pipelines.

In untargeted metabolomics approaches, the inability to structurally annotate relevant features and map them to biochemical pathways is hampering the full exploitation of many metabolomics experiments. Furthermore, variable metabolic content across samples result in sparse feature matrices that are statistically hard to handle. Here, we introduce MS2LDA+ that tackles both above-mentioned problems. Previously, we presented MS2LDA, which extracts biochemically relevant molecular substructures (“Mass2Motifs”) from a collection of fragmentation spectra as sets of co-occurring molecular fragments and neutral losses, thereby recognizing building blocks of metabolomics. Here, we extend MS2LDA to handle multiple metabolomics experiments in one analysis, resulting in MS2LDA+. By linking Mass2Motifs across samples, we expose the variability in prevalence of structurally related metabolite families. We validate the differential prevalence of substructures between two distinct samples groups and apply it to fecal samples. Subsequently, within one sample group of urines, we rank the Mass2Motifs based on their variance to assess whether xenobiotic-derived substructures are among the most-variant Mass2Motifs. Indeed, we could ascribe 22 out of the 30 most-variant Mass2Motifs to xenobiotic-derived substructures including paracetamol/acetaminophen mercapturate and dimethylpyrogallol. In total, we structurally characterized 101 Mass2Motifs with biochemically or chemically relevant substructures. Finally, we combined the discovered metabolite families with full scan feature intensity information to obtain insight into core metabolites present in most samples and rare metabolites present in small subsets now linked through their common substructures. We conclude that by biochemical grouping of metabolites across samples MS2LDA+ aids in structural annotation of metabolites and guides prioritization of analysis by using Mass2Motif prevalence.