Evaluating the Accuracy of the QCEIMS Approach for Computational Prediction of Electron Ionization Mass Spectra of Purines and Pyrimidines

Abstract

Mass spectrometry is the most commonly used method for compound annotation in metabolomics. However, most mass spectra in untargeted assays cannot be annotated with specific compound structures because reference mass spectral libraries are far smaller than the complement of known molecules. Theoretically predicted mass spectra might be used as a substitute for experimental spectra especially for compounds that are not commercially available. For example, the Quantum Chemistry Electron Ionization Mass Spectra (QCEIMS) method can predict 70 eV electron ionization mass spectra from any given input molecular structure. In this work, we investigated the accuracy of QCEIMS predictions of electron ionization (EI) mass spectra for 80 purine and pyrimidine derivatives in comparison to experimental data in the NIST 17 database. Similarity scores between every pair of predicted and experimental spectra revealed that 45% of the compounds were found as the correct top hit when QCEIMS predicted spectra were matched against the NIST17 library of >267,000 EI spectra, and 74% of the compounds were found within the top 10 hits. We then investigated the impact of matching, missing, and additional fragment ions in predicted EI mass spectra versus ion abundances in MS similarity scores. We further include detailed studies of fragmentation pathways such as retro Diels–Alder reactions to predict neutral losses of (iso)cyanic acid, hydrogen cyanide, or cyanamide in the mass spectra of purines and pyrimidines. We describe how trends in prediction accuracy correlate with the chemistry of the input compounds to better understand how mechanisms of QCEIMS predictions could be improved in future developments. We conclude that QCEIMS is useful for generating large-scale predicted mass spectral libraries for identification of compounds that are absent from experimental libraries and that are not commercially available.

Good practices and recommendations for using and benchmarking computational metabolomics metabolite annotation tools

“…While DL developments in metabolomics are still in their infancy, there is reason to be optimistic about their future in the field. In the light of i) current advances in related fields that also look promising (i.e., considering DeepDIA and DLEAMSE developed for proteomics (Qin et al, 2021;Yang et al, 2020), ii) the ever-increasing knowledge of how small molecules behave in the mass spectrometer (i.e., through quantum mechanics calculations (Lee et al, 2022), and iii) the increasing amount of training data, it is very likely that deep learning approaches will substantially boost the field. However, it is unlikely we will arrive there within the next 5-10 years.…”

Calculation of Mass Spectra with the QCxMS Method for Negatively and Multiply Charged Molecules

“…Based on these ideas, the Quantum Chemical Mass Spectrometry program (QCxMS) was developed, which can operate in x = electron ionization (EI), dissociative electron attachment (DEA) and collision-induced dissociation (CID) run modes. The effectiveness of QCxMS to successfully generate in silico spectra in its EI mode is well documented − and has been demonstrated recently by its use for extension of mass spectra databases. − Detailed fragmentation pattern analysis using the EI, DEA, and positive ion CID modes have successfully been conducted earlier. ,− In this work, an extension of the CID run mode is presented, in which the charge state of the molecular ion is no longer restricted to single positive values so that computations of negatively and multiply charged molecular ions are now possible. This improvement is important because common experimental ionization techniques used in tandem with CID − can produce ions with multiple positive or negative charges. , The new charge unrestricted CID mode was tested on a benchmark set of molecules, for which the most apparent fragmentation pathways are discussed in detail.…”

“…The effectiveness of QCxMS to successfully generate in silico spectra in its EI mode is well documented 19−22 and has been demonstrated recently by its use for extension of mass spectra databases. 23 ion CID modes have successfully been conducted earlier. 17,26−28 In this work, an extension of the CID run mode is presented, in which the charge state of the molecular ion is no longer restricted to single positive values so that computations of negatively and multiply charged molecular ions are now possible.…”

Comparative Evaluation of Electron Ionization Mass Spectral Prediction Methods

“…Ongoing improvements of MS prediction algorithms will increase the use of MS prediction methods as sketched in Figure . QCEIMS has been evaluated in recent studies, − and a modification using excited-state molecular dynamics has been proposed by Wang et al The development of CFM-ID is also ongoing, and will likely lead to an improved version of CFM-EI. ,, Recently published, new approaches , based on ML using graph neural networks already claim to outperform NEIMS.…”