Predicting EI+ mass spectra with QCxMS

Contributors

Helge Hecht Wudmir Rojas Zargham Ahmad Julia Jakiela

qcxms-spectra-predictions

Motivation

.pull-left[

• 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. ]

.pull-right[ ]

Method

• Calculations can be performed using multiple quantum chemistry frameworks.
• This includes full ab initio molecular dynamics or parameterized semi-empirical quantum mechanical methods.
• Semi-empirical methods are significantly faster (hours vs. weeks) while yielding reasonable accuracy.

HPC Workflow

• The HPC-based workflow is configured to work with a PBS job scheduler and is therefore somwhat limited in flexibility and user friendliness.
• Due to the difficulties of working with HPC clusters, we ported this workflow to Galaxy.

Galaxy Workflow

• We use xTB to optimize the input geometry.
• The QCxMS neutral run tool performs the equilibration and sampling to create the trajectories.
• The QCxMS production run tool performs the simulation of each trajectory.
• The QCxMS getres tool collects the results into an MSP file.

Galaxy Tool Structure

• Each trajectory is computed in a single job, using data-level parallelism.
• All QCxMS tools utilize the same docker container with an executable which was compiled with an optimized compiler.

Runtime Performance Metrics

• Runtime depends primarily on size of the molecule.
• Structure also plays a crucial role - the more complex the geometry and structure, the longer the computations take.
• Memory requirements scale linear with size of the molecule.

References

• Grimme, S. (2013). Towards First Principles Calculation of Electron Impact Mass Spectra of Molecules. Angewandte Chemie International Edition, 52(24), 6306–6312. https://doi.org/10.1002/anie.201300158
• Bauer, C. A., & Grimme, S. (2016). How to Compute Electron Ionization Mass Spectra from First Principles. The Journal of Physical Chemistry A, 120(21), 3755–3766. https://doi.org/10.1021/acs.jpca.6b02907
• Bannwarth, C., Ehlert, S., & Grimme, S. (2019). GFN2-xTB—An Accurate and Broadly Parametrized Self-Consistent Tight-Binding Quantum Chemical Method with Multipole Electrostatics and Density-Dependent Dispersion Contributions. Journal of Chemical Theory and Computation, 15(3), 1652–1671. https://doi.org/10.1021/acs.jctc.8b01176
• Koopman, J., & Grimme, S. (2021). From QCEIMS to QCxMS: A Tool to Routinely Calculate CID Mass Spectra Using Molecular Dynamics. Journal of the American Society for Mass Spectrometry, 32(7), 1735–1751. https://doi.org/10.1021/jasms.1c00098
• Hecht, H., Rojas, W. Y., Ahmad, Z., Křenek, A., Klánová, J., & Price, E. J. (2024). Quantum Chemistry-Based Prediction of Electron Ionization Mass Spectra for Environmental Chemicals. Analytical Chemistry, 96(33), 13652–13662. https://doi.org/10.1021/acs.analchem.4c02589

Thank you!