Identifying a small molecule from its mass spectrum is the primary open problem in computational metabolomics. This is typically cast as information retrieval: an unknown spectrum is matched against spectra predicted computationally from a large database of chemical structures. However, current approaches to spectrum prediction model the output space in ways that force a tradeoff between capturing high resolution mass information and tractable learning. We resolve this tradeoff by casting spectrum prediction as a mapping from an input molecular graph to a probability distribution over chemical formulas. We further discover that a large corpus of mass spectra can be closely approximated using a fixed vocabulary constituting only 2% of all observed formulas. This enables efficient spectrum prediction using an architecture similar to graph classification - GrAFF-MS - achieving significantly lower prediction error and greater retrieval accuracy than previous approaches.

BACKGROUND: Exposure to environmental tobacco smoke (ETS) leads to higher rates of pulmonary diseases and infections in children. To study the biochemical changes that may precede lung diseases, metabolomic effects on fetal and maternal lungs and plasma from rats exposed to ETS were compared to filtered air control animals. Genome- reconstructed metabolic pathways may be used to map and interpret dysregulation in metabolic networks. However, mass spectrometry-based non-targeted metabolomics datasets often comprise many metabolites for which links to enzymatic reactions have not yet been reported. Hence, network visualizations that rely on current biochemical databases are incomplete and also fail to visualize novel, structurally unidentified metabolites. RESULTS: We present a novel approach to integrate biochemical pathway and chemical relationships to map all detected metabolites in network graphs (MetaMapp) using KEGG reactant pair database, Tanimoto chemical and NIST mass spectral similarity scores. In fetal and maternal lungs, and in maternal blood plasma from pregnant rats exposed to environmental tobacco smoke (ETS), 459 unique metabolites comprising 179 structurally identified compounds were detected by gas chromatography time of flight mass spectrometry (GC-TOF MS) and BinBase data processing. MetaMapp graphs in Cytoscape showed much clearer metabolic modularity and complete content visualization compared to conventional biochemical mapping approaches. Cytoscape visualization of differential statistics results using these graphs showed that overall, fetal lung metabolism was more impaired than lungs and blood metabolism in dams. Fetuses from ETS-exposed dams expressed lower lipid and nucleotide levels and higher amounts of energy metabolism intermediates than control animals, indicating lower biosynthetic rates of metabolites for cell division, structural proteins and lipids that are critical for in lung development. CONCLUSIONS: MetaMapp graphs efficiently visualizes mass spectrometry based metabolomics datasets as network graphs in Cytoscape, and highlights metabolic alterations that can be associated with higher rate of pulmonary diseases and infections in children prenatally exposed to ETS. The MetaMapp scripts can be accessed at http://metamapp.fiehnlab.ucdavis.edu.

SUMMARY: Metabolomic publications and databases use different database identifiers or even trivial names which disable queries across databases or between studies. The best way to annotate metabolites is by chemical structures, encoded by the International Chemical Identifier code (InChI) or InChIKey. We have implemented a web-based Chemical Translation Service that performs batch conversions of the most common compound identifiers, including CAS, CHEBI, compound formulas, Human Metabolome Database HMDB, InChI, InChIKey, IUPAC name, KEGG, LipidMaps, PubChem CID+SID, SMILES and chemical synonym names. Batch conversion downloads of 1410 CIDs are performed in 2.5 min. Structures are automatically displayed. IMPLEMENTATION: The software was implemented in Groovy and JAVA, the web frontend was implemented in GRAILS and the database used was PostgreSQL. AVAILABILITY: The source code and an online web interface are freely available. Chemical Translation Service (CTS): http://cts.fiehnlab.ucdavis.edu CONTACT: [email protected]

BACKGROUND: Structure elucidation of unknown small molecules by mass spectrometry is a challenge despite advances in instrumentation. The first crucial step is to obtain correct elemental compositions. In order to automatically constrain the thousands of possible candidate structures, rules need to be developed to select the most likely and chemically correct molecular formulas. RESULTS: An algorithm for filtering molecular formulas is derived from seven heuristic rules: (1) restrictions for the number of elements, (2) LEWIS and SENIOR chemical rules, (3) isotopic patterns, (4) hydrogen/carbon ratios, (5) element ratio of nitrogen, oxygen, phosphor, and sulphur versus carbon, (6) element ratio probabilities and (7) presence of trimethylsilylated compounds. Formulas are ranked according to their isotopic patterns and subsequently constrained by presence in public chemical databases. The seven rules were developed on 68,237 existing molecular formulas and were validated in four experiments. First, 432,968 formulas covering five million PubChem database entries were checked for consistency. Only 0.6% of these compounds did not pass all rules. Next, the rules were shown to effectively reducing the complement all eight billion theoretically possible C, H, N, S, O, P-formulas up to 2000 Da to only 623 million most probable elemental compositions. Thirdly 6,000 pharmaceutical, toxic and natural compounds were selected from DrugBank, TSCA and DNP databases. The correct formulas were retrieved as top hit at 80-99% probability when assuming data acquisition with complete resolution of unique compounds and 5% absolute isotope ratio deviation and 3 ppm mass accuracy. Last, some exemplary compounds were analyzed by Fourier transform ion cyclotron resonance mass spectrometry and by gas chromatography-time of flight mass spectrometry. In each case, the correct formula was ranked as top hit when combining the seven rules with database queries. CONCLUSION: The seven rules enable an automatic exclusion of molecular formulas which are either wrong or which contain unlikely high or low number of elements. The correct molecular formula is assigned with a probability of 98% if the formula exists in a compound database. For truly novel compounds that are not present in databases, the correct formula is found in the first three hits with a probability of 65-81%. Corresponding software and supplemental data are available for downloads from the authors' website.

BACKGROUND: Metabolomic studies are targeted at identifying and quantifying all metabolites in a given biological context. Among the tools used for metabolomic research, mass spectrometry is one of the most powerful tools. However, metabolomics by mass spectrometry always reveals a high number of unknown compounds which complicate in depth mechanistic or biochemical understanding. In principle, mass spectrometry can be utilized within strategies of de novo structure elucidation of small molecules, starting with the computation of the elemental composition of an unknown metabolite using accurate masses with errors <5 ppm (parts per million). However even with very high mass accuracy (<1 ppm) many chemically possible formulae are obtained in higher mass regions. In automatic routines an additional orthogonal filter therefore needs to be applied in order to reduce the number of potential elemental compositions. This report demonstrates the necessity of isotope abundance information by mathematical confirmation of the concept. RESULTS: High mass accuracy (<1 ppm) alone is not enough to exclude enough candidates with complex elemental compositions (C, H, N, S, O, P, and potentially F, Cl, Br and Si). Use of isotopic abundance patterns as a single further constraint removes >95% of false candidates. This orthogonal filter can condense several thousand candidates down to only a small number of molecular formulas. Example calculations for 10, 5, 3, 1 and 0.1 ppm mass accuracy are given. Corresponding software scripts can be downloaded from http://fiehnlab.ucdavis.edu. A comparison of eight chemical databases revealed that PubChem and the Dictionary of Natural Products can be recommended for automatic queries using molecular formulae. CONCLUSION: More than 1.6 million molecular formulae in the range 0-500 Da were generated in an exhaustive manner under strict observation of mathematical and chemical rules. Assuming that ion species are fully resolved (either by chromatography or by high resolution mass spectrometry), we conclude that a mass spectrometer capable of 3 ppm mass accuracy and 2% error for isotopic abundance patterns outperforms mass spectrometers with less than 1 ppm mass accuracy or even hypothetical mass spectrometers with 0.1 ppm mass accuracy that do not include isotope information in the calculation of molecular formulae.