Published July 4, 2022 | Version v1
Poster Open

Quantum chemistry EI Mass Spectrometry database for exposome research

  • 1. RECETOX Centre, Faculty of Science, Masaryk University, Kotlarska 2, Brno 60200, Czech Republic
  • 2. Institute of Computer Science, Masaryk University, Botanická 554/68a, Brno 602 00, Czech Republic


High-resolution electron ionization mass spectrometry (HR-[EI+]-MS) is a powerful technique that enables reliable structural annotation, yet its applicability is hindered by a lack of high-resolution mass spectral libraries. Plus, mass spectra acquired via single quadrupole instruments differ to those acquired on time-of-flight and Orbitrap instruments, leading to lower match scores when comparing to available nominal mass spectra. In addition, many chemicals are not commercially available and experimental libraries have limited chemical diversity.

Alternatively, spectra can be generated in silico, for instance by machine learning or quantum chemistry methods. Typically, machine learning approaches require a wealth of high-quality training data that often limit full applicability. Generation of mass spectra via quantum chemistry calculations has potential since it does not rely on empirical rules or experimental data and can provide insights of fragmentation processes and reactions features.

As case study, we applied the open-source version of the Quantum Chemical Electron Ionization Mass Spectrometry (QCxMS) to calculate mass spectra for coumarin and estragole at standard 70 eV electron ionization (EI). For each molecule, Born-Oppenheimer molecular dynamics at the extended tight-binding semiempirical quantum mechanical GFN2-xTB and PBE0-D3/SV(p) level were carried out to evaluate accuracy of the in silico generated spectra against openly available experimental reference spectra in the RECETOX Exposome HR-[EI+]-MS library.

Both levels of theory computed mass spectra in agreement with their experimental mass spectra counterparts. Therefore, building quantum chemistry mass spectral libraries of reliable accuracy is potentially feasible. However, the computing time using PBE0-D3/SV(p) is over three orders of magnitude higher compared to GFN2-xTB and computational affordability still needs to be addressed. Ongoing work consists of the in-silico generation of spectra for all chemicals with respective high-resolution electron ionization mass spectra available in order to evaluate wider applicability for future prediction of spectra for chemicals without available reference spectra.



Files (843.9 kB)

Name Size Download all
843.9 kB Preview Download