The search for similarities within a collection of MS/MS spectra is a powerful approach to facilitate the identification of new metabolites (Beniddir et al., 2021). mineMS2 implements an innovative strategy to extract frequent fragmentation patterns containing shared m/z differences between subsets of the spectra (Delabrière et al., in preparation). This method is based on:
-
a new representation of spectra as fragmentation graphs of m/z differences
-
an efficient frequent-subgraph mining algorithm to extract patterns
Each pattern is a graph with ion peaks as nodes and m/z differences as edges. These m/z differences can be any difference between the m/z values of two peaks of a spectrum, provided that they are frequent (i.e. detected in at least two spectra). They therefore include not only neutral losses but also m/z differences between ions that belong to distinct fragmentation paths of the precursor, which may also prove specific to the fragmentation of specific molecules. The candidate molecular formulas for each m/z difference are computed to help interpretation (below 200 Da).
mineMS2 patterns are complementary to those of MS2LDA (van der Hooft et al., 2016), the latter consisting of a list of neutral fragments and losses and being obtained using a probabilistic approach. In particular, the structure of mineMS2 patterns in the form of exact graphs (all m/z differences of the pattern are present in all spectra containing this pattern) facilitates their chemical interpretation.
mineMS2 can be further coupled to the GNPS MS/MS molecular networking methodology (Watrous et al., 2012) to focus on patterns that best explain components of the network.
The package can be installed from GitHub with:
#install.packages("devtools")
devtools::install_github("odisce/mineMS2")Two vignettes detail how to compute and explore the fragmentation patterns (mineMS2_main.Rmd) and how to focus on the patterns that best explain components of the molecular network (mineMS2_coupling-to-gnps.Rmd).
The included dataset, which is used in the examples and vignettes, consists of 51 spectra from the untargeted study of the secondary metabolism of Penicillium nordicum (Hautbergue et al., 2019).
mineMS2: Annotation of spectral libraries with exact fragmentation patterns. Alexis Delabrière, Coline Gianfrotta, Sylvain Dechaumet, Annelaure Damont, Thaïs Hautbergue, Pierrick Roger, Emilien Jamin, Olivier Puel, Christophe Junot, François Fenaille and Etienne A. Thévenot.
[email protected], [email protected], and [email protected]
Beniddir,M.A. et al. (2021) Advances in decomposing complex metabolite mixtures using substructure- and network-based computational metabolomics approaches. Nat. Prod. Rep., 38, 1967–1993. DOI:5B10.1039/D1NP00023C.
Hautbergue,T. et al. (2019) Combination of isotope labeling and molecular networking of tandem mass spectrometry data to reveal 69 unknown metabolites produced by penicillium nordicum. Analytical Chemistry, DOI:5B10.1021/acs.analchem.9b01634.
van der Hooft,J.J.J. et al. (2016) Topic modeling for untargeted substructure exploration in metabolomics. Proceedings of the National Academy of Sciences, 113, 13738–13743. DOI:B10.1073/pnas.1608041113.
Watrous,J. et al. (2012) Mass spectral molecular networking of living microbial colonies. Proceedings of the National Academy of Sciences, 109, E1743–E1752. DOI:10.1073/pnas.1203689109.