2 of 9

Motivation

MS data annotation poses a universal bottleneck in research.
In silico spectra prediction using machine learning or quantum chemistry is a promising technique for annotation of unknown compounds.
QCxMS offers reasonably accurate in silico annotation, especially for organic molecules.
The complexity of quantum chemistry predictions presents challenges for non-HPC experts.
Integrating QCxMS into Galaxy provides valuable molecular insights.

Our Goal: Make semi-empirical Quantum Chemistry (QC)-based predictions accessible without advanced computational skills.

DOI: 10.1093/exposome/osac007

90 analytes spiked in serum

18 ppb

61 annotated

54% annotated wrongly at Level 2

60% annotated wrongly at Level 3

3 of 9

QCxMS Spectra Prediction - Method

Zenodo. https://doi.org/10.5281/zenodo.6791061

Sampling:

Molecular dynamics (MD):
T=500 K
Microcanonical NVE assemble

Ionization and Heating phase

Remove one electron
Internal conversion (IC)
Internal Excess Energy (IEE)

Counting:

50-100 counts in base peak.
Record neutral loss.

Evolution of ion:

MD on ion at steps of 0.5 fs
Track secondary fragmentations recursively.
Choose largest charge part post-dissociation.
Start new trajectories without extra heating.

Electron Impact process

4 of 9

QCxMS Spectra Prediction - HPC Workflow

Hecht, et. al. Quantum chemistry based prediction of electron ionization mass spectra for environmental chemicals. ChemRxiv. 1 0.26434/chemrxiv-2024-2ngwq-v2

HPC workflow: https://doi.org/10.5281/zenodo.10853686

5 of 9

QCxMS Spectra Prediction - Galaxy Workflow

tool parameter

Creating and Structuring files

Molecular optimization

QCxMS spectral prediction

MSP generation

Galaxy workflow: https://doi.org/10.48546/WORKFLOWHUB.WORKFLOW.897.1

6 of 9

QCxMS Galaxy Tool Structure

7 of 9

Runtime Performance Metrics

Slots: 155

Job Runtime (s): 34624

CPU usage time (s): 2325007517

CPU user time (s): 1716160059

CPU system time (s): 608847386

Memory allocated (TB): 0.58

33 atoms

Elements: C, H, O, N, Cl

22 atoms

Elements: C, Cl

6 atoms

Elements: C, H

24 atoms

Elements: C, H, O

Enilconazole

Mirex

Ethylene

Benzophenone

Slots: 605

Job Runtime (s): 679037

CPU usage time (s):10185690473

CPU user time (s): 7641539889

CPU system time (s): 2544150283

Memory allocated (TB): 2.25

Slots: 555

Job Runtime (s): 2070941

CPU usage time (s): 9914289674

CPU user time (s): 7506592875

CPU system time (s): 2407696515

Memory allocated (TB): 2.06

Slots: 830

Job Runtime (s): 1720209

CPU usage time (s):13987695689

CPU user time (s): 10219951769

CPU system time (s): 3767743519

Memory allocated (TB): 3.08

8 of 9

Runtime Performance Metrics - Examples

9 of 9

Acknowledgements

Human Exposome Research Group

Jana Klánová –Head

Kapil Mandrah – MSCA fellow

Žiga Tkalec – ERA fellow

Helge Hecht – PhD candidate

Akrem Jbebli – PhD candidate

Hana Seličová – PhD candidate

Thomas Contini – PhD candidate

Biomarker Analytical Laboratories RI

Eliška Benešová - Researcher

Kateřina Coufaliková – Senior researcher

Gabriela Přibyl Dovrtělová - Researcher

Štěpán Koudelka – Senior researcher

Veronika Vidová – Senior researcher

Helge Hecht – Senior researcher

Wudmir Yudy Rojas Verastegui – Senior researcher

Zargham Ahmad – Researcher

Kristína Gömöryová – Senior Researcher

Trace Analytical Laboratories RI

Petra Přibylová & team

Data services

Richard Hůlek & team

EIRENE RI coordination

Jan Ostřížek

Shachar Dvir

RECETOX RI coordination

Petra Růžičková

Operations and management

Martin Životek, Šárka Palátová & team

Partners

Institute of computer science at MU

CERIT-SC

CESNET

ELIXIR-CZ

Galaxy Project

EIRENE consortia

1 of 9