Skip to main content
NCBI home page
ACS Omega logoLink to ACS Omega
. 2022 Mar 8;7(11):9710–9719. doi: 10.1021/acsomega.1c07272

PyFragMS—A Web Tool for the Investigation of the Collision-Induced Fragmentation Pathways

Yury Kostyukevich 1,*, Sergey Sosnin 1, Sergey Osipenko 1, Oxana Kovaleva 1, Lidiia Rumiantseva 1, Albert Kireev 1, Alexander Zherebker 1, Maxim Fedorov 1, Evgeny N Nikolaev 1,*
PMCID: PMC8945079  PMID: 35350354

Abstract

graphic file with name ao1c07272_0009.jpg

Dissociation induced by the accumulation of internal energy via collisions of ions with neutral molecules is one of the most important fragmentation techniques in mass spectrometry (MS), and the identification of small singly charged molecules is based mainly on the consideration of the fragmentation spectrum. Many research studies have been dedicated to the creation of databases of experimentally measured tandem mass spectrometry (MS/MS) spectra (such as MzCloud, Metlin, etc.) and developing software for predicting MS/MS fragments in silico from the molecular structure (such as MetFrag, CFM-ID, CSI:FingerID, etc.). However, the fragmentation mechanisms and pathways are still not fully understood. One of the limiting obstacles is that protomers (positive ions protonated at different sites) produce different fragmentation spectra, and these spectra overlap in the case of the presence of different protomers. Here, we are proposing to use a combination of two powerful approaches: computing fragmentation trees that carry information of all consecutive fragmentations and consideration of the MS/MS data of isotopically labeled compounds. We have created PyFragMS—a web tool consisting of a database of annotated MS/MS spectra of isotopically labeled molecules (after H/D and/or 16O/18O exchange) and a collection of instruments for computing fragmentation trees for an arbitrary molecule. Using PyFragMS, we investigated how the site of protonation influences the fragmentation pathway for small molecules. Also, PyFragMS offers capabilities for performing database search when MS/MS data of the isotopically labeled compounds are taken into account.

Introduction

Liquid chromatography–tandem mass spectrometry (LC–MS/MS) has become a key technique for the modern high-throughput omics technologies,1−4 and currently, it is the method of choice for screening drugs, pesticides, and metabolites in complex biological mixtures. According to the regulation rules, a compound can be considered identified if the measured chromatographic elution time, accurate mass, and fragmentation spectrum match those obtained for the pure chemical standard of this compound using the same equipment and experimental conditions.5 The rapid accumulation of fragmentation spectra began in 1950s,6 and currently, there are extensive databases of electron ionization (such as NIST or Wiley) and collision-induced dissociation spectra including MzCloud7 (https://www.mzcloud.org/), Metlin,8−10 NIST, and several others.11−13 However, even now, the size of such databases is insufficient. For example, the most recent release NIST 20 contains tandem data for ∼30,000 compounds.14 At the same time, there are more than 100,000,000 described molecules (PubChem) and many more chemically possible compounds (GDB-17 contains 166 billion compounds).15

Another problem that is rarely paid attention to is that during ionization, gas-phase isomeric ions can be formed, which differ only by the location of the site of protonation or deprotonation. For the positive ion electrospray ionization (pos-ESI) mode, such ions are called protomers, and they can be separated only using ion mobility spectrometry.16,17 Different protomers can be obtained by varying the solvent composition18 or by ion–molecule reactions.19 It is known that the MS/MS spectra of different protomers can be remarkably different. Fragmentation spectra available in the databases mentioned above are in fact the overlap of fragmentation spectra corresponding to different protomers. In addition, the ionizing proton can migrate during the excitation. This is especially important when considering the fragmentation of a peptide using the mobile proton model.20

Many research studies have been carried out in the field of development of in silico fragmentation tools, which can be used to identify compounds without a reference MS/MS spectrum. Weissberg and Dagan published 68 fragmentation rules for the interpretation and prediction of ESI-MS/MS spectra.21 Kebarle, Futrell, and others used RRKM (Rice–Ramsperger–Kassel–Marcus) modeling and consideration of internal vibrational energy distributions to explain fragmentation of ions.22−26 Many attempts were undertaken to use quantum chemistry for the interpretation of fragmentation spectra.27−32 Other software packages were created for the prediction of fragmentation,33−36 and currently, the most advanced software package is MassFrontier (uses a rule-based approach and expert curation), MS-FINDER37 (uses a set of hydrogen rearrangement rules), MetFrag38 (uses iterative bond cleavage), CFM-ID,39,40 and SIRIUS 4,41,42 which integrates CSI:FingerID36 (uses fragmentation trees). Since experimental MS/MS spectra contain ions formed by several fragmentation events, it is convenient to use a fragmentation tree concept. A fragmentation tree is the directed graph in which two product ions are connected with an edge only if they can be directly attributed to a single fragmentation event. Such a concept is implemented in different manners in MetFrag, MS-FINDER, and CFM-ID. In the case of CSI:FingerID (see Böcker et al.43−47), tree nodes are annotated with the molecular formulas of the fragments, and the edges represent (neutral or radical) losses.

It is obvious that the annotation of MS/MS fragments and even the quality of in silico fragmentation algorithms can be considerably improved by using large data sets of MS/MS spectra of isotopically labeled compounds. Indeed, it was the use of labeled compounds that helped discover intramolecular rearrangements such as the McLafferty rearrangement,48 formation of the tropylium ion,49 and so forth.50−52 Unfortunately, there are no such public data sets (due to the high cost of isotopically labeled compounds), and only in MetFrag was an attempt undertaken to use the MS/MS data of isotopically labeled compounds for improving the quality of annotation and prediction.53 In addition, no software can account for different sites of ionization.

Recently, we demonstrated a cheap and simple approach for the combination of isotope exchange reactions (H/D and 16O/18O) with high-resolution MS54−56 in order to improve the reliability of compound identification.54,57 Under the proposed conditions, the number of exchanges can be predicted if the structure of the molecule is known.58−61

The isotope exchange reaction combined with high-resolution MS is also a valuable analytical tool for the identification of unknowns. Indeed, the number of exchanges (H/D and 16O/18O) determined experimentally for an unknown molecule can serve as an additional structural descriptor during the database search (by filtering by the number of groups such as −OH, −COOH, =O, and so forth). We recently showed that this can reduce the number of candidates by a factor of 10.54 Additional improvement of the accuracy of identification can be achieved by considering the MS/MS spectra of the compounds after isotope exchange. Recently, we demonstrated that for the given MS/MS spectrum of an unknown molecule after the 16O/18O exchange reaction, the number of possible candidates with a known structure but unknown MS/MS spectra can be reduced via the substructure mapping based on the molecular formula of MS/MS fragments and predicted position of 18O labels.62,63 It is obvious that the accurate consideration of the fragmentation pathways may improve the quality of the identification even more.

Here, we report PyFragMS—a web tool consisting of a database of annotated MS/MS spectra of isotopically labeled molecules (after H/D and 16O/18O exchange), tools and algorithms for the creation of the fragmentation tree for an arbitrary molecule (including isotopically labeled compounds), and tools for the identification of molecules based on the experimentally measured MS/MS spectra after the isotope exchange reaction. We demonstrate how it can assist in the annotation of MS/MS product ions, investigation of the fragmentation pathways, and identification of the unknown.

Methods

Samples

All drug standards were provided by the Central Toxicological laboratory of Russian Ministry of Health.

Data Acquisition and Processing

Mass spectra were acquired on a QExactive Orbitrap system (Thermo Fischer Scientific) with a modified matrix-assisted laser desorption/ionization/ESI injector (Spectroglyph, LLC), operated in the ESI mode at a 140,000 resolution both for MS and MS/MS measurements. Solutions of target compounds in a methanol–water mixture (1:1) were infused with the flow rate of 1 μL·min–1 and the spray voltage of 3 kV. Precursor ions corresponding to protonated (or deuterated) molecules and the isotope exchange reaction products were isolated with a 0.4 Da window and fragmented with various collision energy values ranging from 10 to 90 NCE. Nitrogen was used as the collision gas. Also, for some compounds, an LC–MS/MS system was used (see the Supporting Information for more details). All spectra were preprocessed using XCalibur 4.1 software (Thermo Fischer Scientific) to extract MS/MS spectra for all target ions averaged by different collision energies.

Isotope Exchange Reaction

Hydrogen/deuterium (H/D) exchange was performed in the ESI source using our previously developed approach.54 The vapors of D2O were infused into the desolvation capillary heated to 300 °C. H/D exchange occurs both in the gas phase and due to the moisture penetration into the droplets64 in the liquid phase. The details of the ion source design and its application to investigate various compounds can be found in our previous publications.65 To perform oxygen 16O/18O exchange, we followed the procedure described by Samuel and Silver.66 The target compound was dissolved in H218O; the solution was placed in a sealed glass vial and heated at 95 °C for 15 h.

PyFragMS front-end was developed using Anvil (https://anvil.works/). Anvil allows us to build web apps using Python. PyFragMS back-end was developed using Python. Operations with molecules were realized using the RDKit library (https://www.rdkit.org/) and “molmass” package. The PyFragMS web interface is described in the Supporting Information. For plotting of fragmentation, GraphViz software was used.

The PyFragMS database includes the largest public data set of MS/MS spectra,67 previously published MS/MS spectra of compounds after H/D exchange,53 and data acquired by us (MS/MS spectra of compounds after H/D and 16O/18O exchange). Assigning of molecular formulas to fragment ions was carried out using Sirius 4 software.41 Currently, the database contains MS/MS spectra for >5000 molecules, including spectra for >1000 molecules after H/D exchange and spectra for >100 molecules after the 16O/18O exchange reaction. The majority of the spectra correspond to pos-ESI, and for >200 compounds, the negative ion ESI (neg-ESI) spectra are available. For building fragmentation trees, users can use the data stored in our database or upload their own data. If any of the readers wish to contribute their data to the PyFragMS database, we encourage you to contact the authors.

The PyFragMS interface is shown in Figure 1, and the PyFragMS algorithm to compute the fragmentation tree is shown in Figure 2.

Figure 1.

Figure 1

Open in a new tab

Interface to PyFragMS (https://pyfragms.anvil.app/). The .DataBase block is the interface to the embedded library of experimentally obtained MS/MS spectra after the isotope exchange reaction (H/D and 16O/18O). The .Discovery block is the search tool for the molecular identification based on the experimentally measured MS/MS spectrum, which can be operated using the MS/MS spectra of compounds after the isotope exchange reaction. The .Fragmentation trees block is the tool for the calculation of the fragmentation tree for a molecule, which can account for different sites of protonation and can work with compounds after the isotope exchange reaction.

Figure 2.

Figure 2

Open in a new tab

Description of the algorithm used for the computation of the fragmentation tree.

The PyFragMS approach to compound identification based on the MS/MS similarity search utilizes conventional cosine similarity and Jaccard similarity measures. However, it uses assigned molecular formulas instead of measured m/z. In order to reduce the search space, filtration based on the precursor formula and the number of H/D and 16O/18O exchanges is available.

For the convenience of users, we have recorded a video tutorial explaining the architecture of PyFragMS and step-by-step instructions on how to use it. The tutorial is available at the PyFragMS website (https://pyfragms.anvil.app/).

Step-by-step instructions for using PyFragMS (with screenshots of the interface) are available in the Supporting Information.

Results and Discussion

Database of Isotopically Labeled Compounds

Compounds labeled with stable isotopes have found wide application in MS-based studies. However, as they are primarily used for purposes of quantitative analysis, these compounds are designed to carry isotope labels in a stable part of a molecule (13C, 15N, or deuterium in −C–H covalent bonds). It provides a stable isotope composition during long-term storage and use. The synthesis of such isotope analogues is an expensive and complicated process68 that limits the use of isotopically labeled compounds to study fragmentation and for untargeted identification.69 For example, the MzCloud database contains a very limited number of deuterated molecules; most of them carry the heavy isotope in the terminal methylene groups, which reduces their usefulness for improving the annotation of MS/MS fragment ions.

Recently, we proposed an ion source and a method to perform H/D exchange in the inlet desolvation capillary of a mass spectrometer (see Figure 3A). This technique allows for the counting of labile hydrogen atoms. Such information can enhance MS identification of chemical compounds. When ions with isotope labels produced from the H/D exchange reaction are fragmented in the collision cell of a mass spectrometer, additional information about the molecular structure may be gained from the resulting MS/MS spectra. In-source H/D exchange and in-solution 16O/18O exchange provide an easy and cheap way to collect an MS/MS spectra database of isotopically labeled compounds with a diverse distribution of the label across the molecular structure. Such a database may be used to investigate fragmentation, improve fragment annotation, or predict fragmentation spectra, which is important for untargeted identification workflows. In PyFragMS, we accumulated previously published MS/MS spectra of compounds after H/D exchange53 and the data acquired by us (MS/MS spectra of compounds after H/D and 16O/18O exchange).

Figure 3.

Figure 3

Open in a new tab

(A) Experimental setup used for performing in-ESI source H/D exchange and in-solution 16O/18O exchange. (B1) Experimentally measured MS2 spectrum of MDPV. (B2) Experimental spectrum for MDPV after H/D exchange. (B3) Experimental spectrum for 18O-labeled MDPV. (C) Structure of the MDPV molecule, showing the site of 16O/18O exchange and two possible sites of protonation (deuteration). (D1,D2) Computed fragmentation trees for different protonation sites. (E) Correct fragmentation pathway and structure for the 135 ion proved using MSn experiments, DFT calculations, and infrared ion spectroscopy.70 (F) Annotation of fragment ions proposed by MzCloud and MetFrag. The selected m/z = 135.05 was annotated wrong.

Computing Fragmentation Trees

We will demonstrate the developed approach by computing fragmentation trees and determining the correct fragmentation pathways for methylenedioxypyrovalerone (MDPV), a psychoactive designer drug. The acquired MS/MS spectra in the pos-ESI mode for MDPV, MDPV after H/D exchange, and MDPV after 16O/18O exchange are shown in Figure 3B. We can see that certain oxygen containing fragments carry an 18O isotope; also, we can see that only two fragments (m/z = 234 and m/z = 84) carry deuterium (note that we are using only integer values of ions in the text). Assigning the correct number of exchanges for each fragment was performed based on the accurate mass difference for H/D exchange (1.006277 Da) and for 16O/18O exchange (2.004245 Da). Despite the fact that neutral MDPV does not have protogenic groups, the proton attaching during ionization in the pos-ESI mode is labile and is exchanged for deuterium. It is worth noting that we may not observe intermediate short-lived metastable ions.

When investigating the fragmentation pathways of any molecule, we must take into account the possible coexistence of several protomers (Figure 3C). The computed fragmentation trees for MDPV protonated at two different sites are shown in Figure 3D. The proposed algorithm was able to assign structures to most fragment peaks (seven out of eight), observed in the experimental MS/MS spectrum of MDPV. The use of isotope labels allowed considerable reduction of the number of possible candidate structures. We can see that it is possible to propose structures for m/z 233 for both protonation sites; however, m/z 126 can be formed only when the C=O group is protonated. Accordingly, m/z values of 205, 175, 149, and 84 correspond to the protonation of nitrogen. One peak with m/z 135 was not annotated. In Figure 3E, the correct pathway that leads to this fragment formation is shown, as was described by Davidson.70 This pathway was proven by authors using MSn experiments, DFT calculations, and infrared ion spectroscopy. Unfortunately, currently, our algorithm cannot account for possible intramolecular rearrangement. In Figure 3F, we show the annotation of fragment ions performed using MzCloud and using MetFrag. It can be seen that both MzCloud and MetFrag incorrectly annotated m/z 135. The proposed annotation is wrong because such a fragment should carry an 18O label after 16O/18O exchange; however, in the experimental spectrum, there is no such label.

Note that when computing a fragmentation tree, PyFragMS allows the elemental composition of the theoretical structure to differ from that experimentally measured by one hydrogen. This allows for the possible hydrogen transfer during the fragmentation. Also, PyFragMS produces a table output showing for each MS/MS ion the IDs of the corresponding nodes in the fragmentation tree and IDs of nodes for which experimental and theoretical formulas coincide or differ. The decision on which nodes remain should be made by the user.

There are many molecules that produce protomers in the ESI source, and it often happens that there can be three or even more possible sites of protonation. Generally, the bigger the molecule is, the more the possible sites of protonation are. In Figure 4, we demonstrate the automatically generated fragmentation tree for a bisacodyl molecule considering three different sites of protonation. It is remarkable that when nitrogen is protonated, then structures are found for only one fragment (out of six). Also, the structures of m/z 319 are different depending on which oxygen in the ester group is protonated. When the C=O group is protonated, m/z 319 can be further fragmented to yield only m/z 183, while when −O– is protonated, one of the possible structures of m/z 319 can yield m/z 226, which further produces m/z 183, 154, 199, and 166.

Figure 4.

Figure 4

Open in a new tab

Automatically generated fragmentation tree for the bisacodyl molecule with three different sites of protonation (marked with an arrow). (A–C) Trees for different protonation sites, (D) fragment ions, and (E) structure of the bisacodyl molecule.

Since protomers produce different fragment ions during the fragmentation process, it is important to accumulate a set of fragmentation trees in order to understand which fragmentation pathway is favored. We started computing and analyzing fragmentation trees for compounds with only one exchangeable proton (the one attached during ionization). Because the fragmentation is a multistep process, we focused on annotating the fragments resulting from a single fragmentation event. Our results for several selected molecules are summarized in Figure 5 (more data can be found in the Supporting Information). We can see that deuterium seldom appears in the fragment ions. Previously, many research studies have been dedicated to the role of protonation on the electron density rearrangement in the molecule and on the bond dissociation.71 The bond weakening induced by the protonation may be explained by the tendency of a protonated atom to recover its electroneutrality by lowering the electron density along bonds with neighbor atoms. As a result of fragmentation, the fragment with a proton (or deuterium) becomes neutral, and the rest of the molecule becomes positively charged.71 Summarizing the experimental results, we formulated the following rules that explain the majority of the experimental data, not including intramolecular rearrangements:

  • (1)

    A molecule can be protonated at different sites due to lone electron pairs of heteroatoms (N, O, and S).

  • (2)

    The primary fragmentation occurs near the protonation site due to the electron density rearrangement. As a result of fragmentation, the fragment with a proton (or deuterium) is neutral, and the fragment without a proton carries a charge. We observe this in the spectrum.

  • (3)

    Less preferably, the fragmentation can occur in any other place of a molecule; however, the even number of electrons rule must be followed.72 This includes hydrogen rearrangement from neighboring atoms if necessary.

Figure 5.

Figure 5

Open in a new tab

Proposed protonation sites and fragmentation pathways explaining experimental results of the deuterium incorporation in the fragment ions for different molecules. (A) Cinnarizine, (B) climbazole, (C) tolperisone, (D) thioridazine, (E) benzydamine, (F) bisacodyl, (G) anastrozole, and (H) buflomedil.

The pathways shown in Figure 5 generally support these rules. Sometimes, during the fragmentation of the deuterated precursor, we observed the appearance of the labeled fragment ions, which indicates the second protonation (deuteration) site with the comparable abundance (see Figure 6). The peak with deuterium appears via fragmentation near the deuteration site and charge transfer, while the peak with deuterium corresponds to the fragmentation far from the deuteration site and hydrogen transfer to account for the even electron rule.

Figure 6.

Figure 6

Open in a new tab

Explanation of the observation of fragment peaks with and without deuterium when fragmenting the deuterated precursor. If there are several sites of protonation in the molecule, the fragmentation can occur via different mechanisms involving charge transfer to the ionization site (produces a non-deuterated peak) and hydrogen transfer far from the ionization site (produces a deuterated peak).

Identification of the Unknown

PyFragMS offers functionality for the identification of the unknown. PyFragMS allows choosing between conventional cosine and Jaccard similarity measures when performing MS/MS search. In addition, the reduction of the search space is possible based on the number of H/D and 16O/18O of the precursor ion. Because PyFragMS contains the database of MS/MS spectra of isotopically labeled molecules, it is possible to use isotopically labeled fragment ions for MS/MS search. Such an approach considerably improves the reliability of the identification. As a model problem, we considered the identification of MDPV when we measured precursor ions and only one fragment ion (m/z = 149). However, we possess information of the number of labile H and O in the precursor and fragment. In Table 1, we can see how the use of the isotope exchange improves the identification.

Table 1. Application of the Isotope Exchange for Improving the Identification.

input data number of results ID of the correct molecule
fragment ion 19 12
precursor ion, fragment ion 5 1
fragment ion, number of H/D exchanges of the precursor 14 8
fragment ion, number of 16O/18O exchanges of the precursor 6 1
fragment ion, number of 16O/18O exchanges of the fragment 1 1
Open in a new tab

When identifying an unknown molecule for which the MS/MS spectrum is not yet included in any database, the following strategy may be utilized using PyFragMS. For a given MS/MS spectrum, fragmentation trees can be generated for all isomers of the precursor molecule, and based on some score, the resulting molecules can be chosen.

We demonstrate this in Figure 7A. For two isomers of mephedrone, we computed fragmentation trees taking the structure of one molecule and the MS/MS spectrum of another (issue of protomers is omitted for simplicity). We can see that we cannot generate a fragmentation tree for 7-APBD and the MS/MS spectrum of mephedrone. The tree for mephedrone and the MS/MS spectrum of 7-APBD can be generated but requires breaking cyclic structures at the first stage. Trees obtained for molecules and corresponding MS/MS spectra look more reasonable.

Figure 7.

Figure 7

Open in a new tab

(A) Fragmentation trees computed when SMILES was taken from one molecule and experimentally measured fragment ions from another. (B) Table of the proposed quality index for 15 isomers of mephedrone.

We generated fragmentation trees for 15 isomers of mephedrone (corresponding MS/MS spectra are included in PyFragMS) for which MS/MS spectra were available in the MzCloud database, interchanging the precursor molecule and MS/MS spectra. For each combination, we calculated the following index

graphic file with name ao1c07272_m001.jpg

Here, Ninput_frag—the number of the input fragments, Nused_frag—the number of fragments that were used in the computed tree, Nedges—the total number of edges in the tree, and Nbreaking_bonds—the total sum of all breaking bonds in the tree (1 for each breaking single bond, 2 for breaking two bonds etc.). Value in the first parenthesis equals the portion of the fragments used in the computed tree. Value in the second parenthesis is the portion of breaking more than one bond in the single fragmentation event. Our results are presented in Figure 7B. We can see that generally, I is higher when the structure and MS/MS spectrum correspond to the same molecule. However, for some cases (APBD isomers), I = 1, and fragmentation trees look reasonable for all combinations. Such an approach can be used for the identification.

Of course, the proposed approach and formula for calculating the proposed index requires more research, consideration of the fragmentation trees of isomers of different molecules, and so forth. The most challenging question would be choosing the correct protonation sites. The continuing accumulation of the MS/MS spectra of isotopically labeled compounds shall help a lot. We are planning to focus on this in the near future.

Conclusions

We have created PyFragMS—a web tool consisting of a database of annotated MS/MS spectra of isotopically labeled molecules (after H/D and 16O/18O exchange), instruments for creating the fragmentation tree for an arbitrary molecule, and tools for the identification of the unknown using isotope exchange information and MS/MS data. It was demonstrated that a simple isotope exchange experiment performed by using a previously described approach and subsequent consideration of MS/MS data of labeled compounds via computing fragmentation trees improves fragment annotation and allows investigation of fragmentation pathways. Using PyFragMS, it is possible to investigate how the site of protonation influences the fragmentation pathway for small molecules. Currently, we are using PyFragMS to build a large database of correct fragmentation pathways taking into account different possible sites of protonation or deprotonation. We believe that such a database will considerably help improve the quality of the software for predicting MS/MS fragments in silico, and we are inviting researchers working with isotopically labeled molecules to share data that will be included into the PyFragMS database (feel free to contact authors). The use of the MS/MS data of isotopically labeled compounds can considerably increase the reliability of the identification, which is especially important when the precursor ion is not known (SWATH,73 etc.).

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsomega.1c07272.

Author Contributions

The manuscript was written through contributions of all authors. All authors have given approval to the final version of the manuscript.

The research was supported by the Russian Scientific Foundation grant no 18-79-10127.

The authors declare no competing financial interest.

Supplementary Material

ao1c07272_si_001.pdf (1.2MB, pdf)

References

  1. Aebersold R.; Mann M. Mass spectrometry-based proteomics. Nature 2003, 422, 198–207. 10.1038/nature01511. [DOI] [PubMed] [Google Scholar]
  2. Dettmer K.; Aronov P. A.; Hammock B. D. Mass spectrometry-based metabolomics. Mass Spectrom. Rev. 2007, 26, 51–78. 10.1002/mas.20108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Zhang X.; et al. Mass spectrometry-based “omics” technologies in cancer diagnostics. Mass Spectrom. Rev. 2007, 26, 403–431. 10.1002/mas.20132. [DOI] [PubMed] [Google Scholar]
  4. Shevchenko A.; Simons K. Lipidomics: coming to grips with lipid diversity. Nat. Rev. Mol. Cell Biol. 2010, 11, 593–598. 10.1038/nrm2934. [DOI] [PubMed] [Google Scholar]
  5. Blaženović I.; Kind T.; Ji J.; Fiehn O. Software tools and approaches for compound identification of LC-MS/MS data in metabolomics. Metabolites 2018, 8, 31. 10.3390/metabo8020031. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. McLafferty F. W.; Stauffer D. B. Retrieval and interpretative computer programs for mass spectrometry. J. Chem. Inf. Comput. Sci. 1985, 25, 245–252. 10.1021/ci00047a021. [DOI] [Google Scholar]
  7. Wilson C. T. R.; Taylor G. I. The bursting of soap-bubbles in a uniform electric field. Math. Proc. Camb. Phil. Soc. 1925, 22, 728–730. 10.1017/s0305004100009609. [DOI] [Google Scholar]
  8. Smith C. A.; et al. METLIN: a metabolite mass spectral database. Ther. Drug Monit. 2005, 27, 747–751. 10.1097/01.ftd.0000179845.53213.39. [DOI] [PubMed] [Google Scholar]
  9. Domingo-Almenara X.; et al. The METLIN small molecule dataset for machine learning-based retention time prediction. Nat. Commun. 2019, 10, 5811. 10.1038/s41467-019-13680-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Domingo-Almenara X.; et al. Autonomous METLIN-guided in-source fragment annotation for untargeted metabolomics. Anal. Chem. 2019, 91, 3246–3253. 10.1021/acs.analchem.8b03126. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Vinaixa M.; et al. Mass spectral databases for LC/MS-and GC/MS-based metabolomics: State of the field and future prospects. Trac. Trends Anal. Chem. 2016, 78, 23–35. 10.1016/j.trac.2015.09.005. [DOI] [Google Scholar]
  12. Stein S. E.; Scott D. R. Optimization and testing of mass spectral library search algorithms for compound identification. J. Am. Soc. Mass Spectrom. 1994, 5, 859–866. 10.1016/1044-0305(94)87009-8. [DOI] [PubMed] [Google Scholar]
  13. Ausloos P.; et al. The critical evaluation of a comprehensive mass spectral library. J. Am. Soc. Mass Spectrom. 1999, 10, 287–299. 10.1016/s1044-0305(98)00159-7. [DOI] [PubMed] [Google Scholar]
  14. Acceptance Criteria for Confirmation of Identity of Chemical Residues using Exact Mass Data for the FDA FVM Program.
  15. Ruddigkeit L.; Van Deursen R.; Blum L. C.; Reymond J.-L. Enumeration of 166 billion organic small molecules in the chemical universe database GDB-17. J. Chem. Inf. Model. 2012, 52, 2864–2875. 10.1021/ci300415d. [DOI] [PubMed] [Google Scholar]
  16. McCullagh M.; Giles K.; Richardson K.; Stead S.; Palmer M. Investigations into the performance of travelling wave enabled conventional and cyclic ion mobility systems to characterise protomers of fluoroquinolone antibiotic residues. Rapid Commun. Mass Spectrom. 2019, 33, 11–21. 10.1002/rcm.8371. [DOI] [PubMed] [Google Scholar]
  17. Karpas Z.; Berant Z.; Stimac R. M. An ion mobility spectrometry/mass spectrometry (IMS/MS) study of the site of protonation in anilines. Struct. Chem. 1990, 1, 201–204. 10.1007/bf00674262. [DOI] [Google Scholar]
  18. Demireva M.; Armentrout P. B. Relative Energetics of the Gas Phase Protomers of p-Aminobenzoic Acid and the Effect of Protonation Site on Fragmentation. J. Phys. Chem. A 2021, 125, 2849–2865. 10.1021/acs.jpca.0c11540. [DOI] [PubMed] [Google Scholar]
  19. Xia H.; Attygalle A. B. Transformation of the gas-phase favored O-protomer of p-aminobenzoic acid to its unfavored N-protomer by ion activation in the presence of water vapor: A n ion-mobility mass spectrometry study. J. Mass Spectrom. 2018, 53, 353–360. 10.1002/jms.4066. [DOI] [PubMed] [Google Scholar]
  20. Paizs B.; Suhai S. Fragmentation pathways of protonated peptides. Mass Spectrom. Rev. 2005, 24, 508–548. 10.1002/mas.20024. [DOI] [PubMed] [Google Scholar]
  21. Weissberg A.; Dagan S. Interpretation of ESI (+)-MS-MS spectra—Towards the identification of “unknowns”. Int. J. Mass Spectrom. 2011, 299, 158–168. 10.1016/j.ijms.2010.10.024. [DOI] [Google Scholar]
  22. Klassen J. S.; Kebarle P. Collision-induced dissociation threshold energies of protonated glycine, glycinamide, and some related small peptides and peptide amino amides. J. Am. Chem. Soc. 1997, 119, 6552–6563. 10.1021/ja962813m. [DOI] [Google Scholar]
  23. Laskin J.; Futrell J. H. On the efficiency of energy transfer in collisional activation of small peptides. J. Chem. Phys. 2002, 116, 4302–4310. 10.1063/1.1450544. [DOI] [Google Scholar]
  24. Paizs B. l.; Suhai S. n. Combined quantum chemical and RRKM modeling of the main fragmentation pathways of protonated GGG. II. Formation of b2, y1, and y2 ions. Rapid Commun. Mass Spectrom. 2002, 16, 375–389. 10.1002/rcm.586. [DOI] [PubMed] [Google Scholar]
  25. Paizs B. l.; Suhai S. n. Combined quantum chemical and RRKM modeling of the main fragmentation pathways of protonated GGG. I. Cis-trans isomerization around protonated amide bonds. Rapid Commun. Mass Spectrom. 2001, 15, 2307–2323. 10.1002/rcm.507. [DOI] [Google Scholar]
  26. Meroueh S. O.; Wang Y.; Hase W. L. Direct dynamics simulations of collision-and surface-induced dissociation of N-protonated glycine. Shattering fragmentation. J. Phys. Chem. A 2002, 106, 9983–9992. 10.1021/jp020664q. [DOI] [Google Scholar]
  27. Mayer I.; Gömöry Á. Use of energy partitioning for predicting primary mass spectrometric fragmentation steps: A preliminary account. Int. J. Quantum Chem. 1993, 48, 599–605. 10.1002/qua.560480854. [DOI] [Google Scholar]
  28. Ásgeirsson V.; Bauer C. A.; Grimme S. Quantum chemical calculation of electron ionization mass spectra for general organic and inorganic molecules. Chem. Sci. 2017, 8, 4879–4895. 10.1039/c7sc00601b. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Bauer C. A.; Grimme S. How to compute electron ionization mass spectra from first principles. J. Phys. Chem. A 2016, 120, 3755–3766. 10.1021/acs.jpca.6b02907. [DOI] [PubMed] [Google Scholar]
  30. Schüler J.-A.; Neumann S.; Müller-Hannemann M.; Brandt W. ChemFrag: chemically meaningful annotation of fragment ion mass spectra. J. Mass Spectrom. 2018, 53, 1104–1115. 10.1002/jms.4278. [DOI] [PubMed] [Google Scholar]
  31. Schymanski E.; Neumann S. The critical assessment of small molecule identification (CASMI): challenges and solutions. Metabolites 2013, 3, 517–538. 10.3390/metabo3030517. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Koopman J.; Grimme S. From QCEIMS to QCxMS: A Tool to Routinely Calculate CID Mass Spectra Using Molecular Dynamics. J. Am. Soc. Mass Spectrom. 2021, 32, 1735. 10.1021/jasms.1c00098. [DOI] [PubMed] [Google Scholar]
  33. Wang Y.; Kora G.; Bowen B. P.; Pan C. MIDAS: A Database-Searching Algorithm for Metabolite Identification in Metabolomics. Anal. Chem. 2014, 86, 9496–9503. 10.1021/ac5014783. [DOI] [PubMed] [Google Scholar]
  34. Verdegem D.; Lambrechts D.; Carmeliet P.; Ghesquiere B. Improved metabolite identification with MIDAS and MAGMa through MS/MS spectral dataset-driven parameter optimization. Metabolomics 2016, 12, 98. 10.1007/s11306-016-1036-3. [DOI] [Google Scholar]
  35. Meringer M.; Reinker S.; Zhang J. A.; Muller A. MS/MS Data Improves Automated Determination of Molecular Formulas by Mass Spectrometry. MATCH Commun. Math. Comput. Chem. 2011, 65, 259–290. [Google Scholar]
  36. Dührkop K.; Shen H.; Meusel M.; Rousu J.; Böcker S. Searching molecular structure databases with tandem mass spectra using CSI: FingerID. Proc. Natl. Acad. Sci. U.S.A. 2015, 112, 12580–12585. 10.1073/pnas.1509788112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Tsugawa H.; et al. Hydrogen rearrangement rules: computational MS/MS fragmentation and structure elucidation using MS-FINDER software. Anal. Chem. 2016, 88, 7946–7958. 10.1021/acs.analchem.6b00770. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Ruttkies C.; Schymanski E. L.; Wolf S.; Hollender J.; Neumann S. MetFrag relaunched: incorporating strategies beyond in silico fragmentation. J. Cheminf. 2016, 8, 3. 10.1186/s13321-016-0115-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Allen F.; Greiner R.; Wishart D. Competitive fragmentation modeling of ESI-MS/MS spectra for putative metabolite identification. Metabolomics 2015, 11, 98–110. 10.1007/s11306-014-0676-4. [DOI] [Google Scholar]
  40. Djoumbou-Feunang Y.; et al. CFM-ID 3.0: Significantly improved ESI-MS/MS prediction and compound identification. Metabolites 2019, 9, 72. 10.3390/metabo9040072. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Dührkop K.; et al. SIRIUS 4: a rapid tool for turning tandem mass spectra into metabolite structure information. Nat. Methods 2019, 16, 299–302. 10.1038/s41592-019-0344-8. [DOI] [PubMed] [Google Scholar]
  42. Dührkop K.; et al. Systematic classification of unknown metabolites using high-resolution fragmentation mass spectra. Nat. Biotechnol. 2020, 39, 462. 10.1038/s41587-020-0740-8. [DOI] [PubMed] [Google Scholar]
  43. Böcker S.; Rasche F. Towards de novo identification of metabolites by analyzing tandem mass spectra. Bioinformatics 2008, 24, i49–i55. 10.1093/bioinformatics/btn270. [DOI] [PubMed] [Google Scholar]
  44. Böcker S.; Dührkop K. Fragmentation trees reloaded. J. Cheminf. 2016, 8, 5. 10.1186/s13321-016-0116-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Rasche F.; Svatoš A.; Maddula R. K.; Böttcher C.; Böcker S. Computing fragmentation trees from tandem mass spectrometry data. Anal. Chem. 2011, 83, 1243–1251. 10.1021/ac101825k. [DOI] [PubMed] [Google Scholar]
  46. Rasche F.; et al. Identifying the unknowns by aligning fragmentation trees. Anal. Chem. 2012, 84, 3417–3426. 10.1021/ac300304u. [DOI] [PubMed] [Google Scholar]
  47. Vaniya A.; Fiehn O. Using fragmentation trees and mass spectral trees for identifying unknown compounds in metabolomics. Trac. Trends Anal. Chem. 2015, 69, 52–61. 10.1016/j.trac.2015.04.002. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. McLafferty F. W. Mass spectrometric analysis. Molecular rearrangements. Anal. Chem. 1959, 31, 82–87. 10.1021/ac60145a015. [DOI] [Google Scholar]
  49. Meyer F.; Harrison A. G. A mechanism for tropylium ion formation by electron impact. J. Am. Chem. Soc. 1964, 86, 4757–4761. 10.1021/ja01076a004. [DOI] [Google Scholar]
  50. Budzikiewicz H.; Wilson J. M.; Djerassi C. Mass spectrometry in structural and stereochemical problems. XXXII. 1 Pentacyclic triterpenes. J. Am. Chem. Soc. 1963, 85, 3688–3699. 10.1021/ja00905a036. [DOI] [Google Scholar]
  51. Duffield A. M.; Budzikiewicz H.; Djerassi C. Mass Spectrometry in Structural and Stereochemical Problems. LXXII. 1 A Study of the Fragmentation Processes of Some Tobacco Alkaloids2. J. Am. Chem. Soc. 1965, 87, 2926–2932. 10.1021/ja01091a024. [DOI] [PubMed] [Google Scholar]
  52. Senn M.; Richter W. J.; Burlingame A. L. Convenient deuterium labeling for mass spectrometry via exchange of enolizable hydrogen on a gas-liquid chromatography column. J. Am. Chem. Soc. 1965, 87, 680–681. 10.1021/ja01081a069. [DOI] [Google Scholar]
  53. Ruttkies C.; et al. Supporting non-target identification by adding hydrogen deuterium exchange MS/MS capabilities to MetFrag. Anal. Bioanal. Chem. 2019, 411, 4683–4700. 10.1007/s00216-019-01885-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Kostyukevich Y.; et al. Hydrogen/Deuterium and 16O/18O-Exchange Mass Spectrometry Boosting the Reliability of Compound Identification. Anal. Chem. 2020, 92, 6877–6885. 10.1021/acs.analchem.9b05379. [DOI] [PubMed] [Google Scholar]
  55. Kostyukevich Y.; et al. Hydrogen/Deuterium Exchange Aiding Compound Identification for LC-MS and MALDI Imaging Lipidomics. Anal. Chem. 2019, 91, 13465–13474. 10.1021/acs.analchem.9b02461. [DOI] [PubMed] [Google Scholar]
  56. Kostyukevich Y.; et al. Hydrogen/deuterium exchange in mass spectrometry. Mass Spectrom. Rev. 2018, 37, 811–853. 10.1002/mas.21565. [DOI] [PubMed] [Google Scholar]
  57. Budzikiewicz H. Mass spectrometry in natural product structure elucidation. Prog. Chem. Org. Nat. Prod. 2015, 100, 77–221. 10.1007/978-3-319-05275-5_2. [DOI] [PubMed] [Google Scholar]
  58. Zherebker A. Y.; et al. Refinement of compound aromaticity in complex organic mixtures by stable isotope label assisted ultra-high resolution mass spectrometry. Anal. Chem. 2020, 92, 9032. 10.1021/acs.analchem.0c01208. [DOI] [PubMed] [Google Scholar]
  59. Kostyukevich Y.; Kononikhin A.; Popov I.; Nikolaev E. Analytical Description of the H/D Exchange Kinetic of Macromolecule. Anal. Chem. 2018, 90, 5116–5121. 10.1021/acs.analchem.7b05151. [DOI] [PubMed] [Google Scholar]
  60. Acter T.; et al. Optimization and application of APCI hydrogen–deuterium exchange mass spectrometry (HDX MS) for the speciation of nitrogen compounds. J. Am. Soc. Mass Spectrom. 2015, 26, 1522–1531. 10.1007/s13361-015-1166-2. [DOI] [PubMed] [Google Scholar]
  61. Cho Y.; Ahmed A.; Kim S. Application of Atmospheric Pressure Photo Ionization Hydrogen/Deuterium Exchange High-Resolution Mass Spectrometry for the Molecular Level Speciation of Nitrogen Compounds in Heavy Crude Oils. Anal. Chem. 2013, 85, 9758–9763. 10.1021/ac402157r. [DOI] [PubMed] [Google Scholar]
  62. Rumiantseva L.; et al. Increasing the reliability of compound identification in biological samples using 16O/18O-exchange mass spectrometry. Anal. Bioanal. Chem. 2022, 414, 2537. 10.1007/s00216-022-03924-9. [DOI] [PubMed] [Google Scholar]
  63. Osipenko S.; et al. Oxygen Isotope Exchange Reaction for Untargeted LC–MS Analysis. J. Am. Soc. Mass Spectrom. 2022, 33, 390–398. 10.1021/jasms.1c00383. [DOI] [PubMed] [Google Scholar]
  64. Wang R.; Zenobi R. Evolution of the solvent polarity in an electrospray plume. J. Am. Soc. Mass Spectrom. 2010, 21, 378–385. 10.1016/j.jasms.2009.10.022. [DOI] [PubMed] [Google Scholar]
  65. Kostyukevich Y.; et al. Hydrogen/Deuterium Exchange Aiding Compound Identification for LC-MS and MALDI Imaging Lipidomics. Anal. Chem. 2019, 91, 13465. 10.1021/acs.analchem.9b02461. [DOI] [PubMed] [Google Scholar]
  66. Samuel D.; Silver B. L.. Oxygen Isotope Exchange Reactions of Organic Compounds. In Advances in Physical Organic Chemistry; Gold V., Ed.; Academic Press, 1965; pp 123–186. [Google Scholar]
  67. Tsugawa H.; et al. MS-DIAL: data-independent MS/MS deconvolution for comprehensive metabolome analysis. Nat. Methods 2015, 12, 523–526. 10.1038/nmeth.3393. [DOI] [PMC free article] [PubMed] [Google Scholar]
  68. Atzrodt J.; Derdau V.; Fey T.; Zimmermann J. The renaissance of H/D exchange. Angew. Chem., Int. Ed. 2007, 46, 7744–7765. 10.1002/anie.200700039. [DOI] [PubMed] [Google Scholar]
  69. Tsugawa H.; et al. A cheminformatics approach to characterize metabolomes in stable-isotope-labeled organisms. Nat. Methods 2019, 16, 295–298. 10.1038/s41592-019-0358-2. [DOI] [PubMed] [Google Scholar]
  70. Davidson J. T.Structural Characterization of Emerging Synthetic Drugs. Ph.D. Thesis, Graduate Theses, Dissertations, and Problem Reports, 7584; Eberly College of Arts and Sciences, 2020. [Google Scholar]
  71. Bouchoux G. Gas phase basicities of polyfunctional molecules. Part 6: Cyanides and isocyanides. Mass Spectrom. Rev. 2018, 37, 533–564. 10.1002/mas.21538. [DOI] [PubMed] [Google Scholar]
  72. Karni M.; Mandelbaum A. The ‘even-electron rule. Org. Mass Spectrom. 1980, 15, 53–64. 10.1002/oms.1210150202. [DOI] [Google Scholar]
  73. Collins B. C.; et al. Multi-laboratory assessment of reproducibility, qualitative and quantitative performance of SWATH-mass spectrometry. Nat. Commun. 2017, 8, 291–12. 10.1038/s41467-017-00249-5. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

ao1c07272_si_001.pdf (1.2MB, pdf)

Articles from ACS Omega are provided here courtesy of American Chemical Society

RESOURCES