Molecular structure elucidation by integrating different data mining strategies

January 4, 2019
In Progress
Share this project


The overall goal of this project is to develop and implement advanced data-driven programming tools, enabling a superior insight into ultra-high performance liquid chromatography coupled to high- resolution mass spectrometry (UHPLC-HRMS) data. While HPLC has been used as the first level of analyte separation since the 1960s, HRMS is a relatively new and powerful analytic technique used for discovery of molecular species based on their exact mass to charge ratio (m/z). The instrumentation applied is capable of separating mass fragments at the fourth or fifth decimal place. The additional information narrows down the possible chemical formulas of a molecule and thus allows an unprecedented unambiguous qualitative and quantitative assessment of the composition of various types of samples. Not surprisingly, HRMS has found applications across a broad spectrum of scientific fields.

Although we can routinely discern hundreds to thousands of molecular ‘features’ in complex samples such as blood, aerosols, soil, or biofuels, the complexity of the resulting data stream increases proportionally, producing millions of data points per second in multidimensional space. Thus post-processing and data reduction methods followed by data mining and innovative visualization techniques are required to yield meaningful information from HRMS. The project is about developing semi-automatic methods to confidently pinpoint each unknown molecular structure. It is a unique opportunity to expand the applicability of both HRMS and the Kendrick Mass Defect (KMD) approach beyond their current state-of-the-art applications, as well as beyond the capabilities of other analytic methods such as NMR and X-ray crystallography tools that typically require pure samples in relatively large amounts.



SDSC Team:
Eliza Harris
Fernando Perez-Cruz
Guillaume Obozinski
Lilian Gasser
Michele Volpi

PI | Partners:

Catalytic Process Engineering Research Group:

  • Dr. Saša Bjelić

More info


  • Molecular clustering based on UHPLC-HRMS/MS data reflecting chemical “families” based on the presence of similar functional groups.
  • Within-cluster prediction of functional groups and molecular structure for unknown compounds.
  • Predictive modelling of molecular fragmentation patterns, retention time, and other features.
Figure 1: Fragmentation spectra for two dicarboxylic acids illustrating clear differences in fragment patterns and intensities despite similar structures.



Additionnal resources


  1. Wu et al. (2021) Valence Photoionization and Energetics of Vanillin, a Sustainable Feedstock Candidate, The Journal of Physical Chemistry A, doi: 10.1021/acs.jpca.1c00876
  2. Dührkop et al. (2020) Systematic classification of unknown metabolites using high-resolution fragmentation mass spectra, Nature Biotechnology, doi: 10.1038/s41587-020-0740-8
  3. Arturi et al. (2019) Molecular footprint of co-solvents in hydrothermal liquefaction (HTL) of Fallopia Japonica, Journal of Supercritical Fluids, doi: 10.1016/j.supflu.2018.08.010
  4. Roach et al. (2011) Higher-Order Mass Defect Analysis for Mass Spectra of Complex Organic Mixtures, Analytical Chemistry, doi: 10.1021/ac200654j


Related Pages

More projects


In Progress
Machine Learning for the Future Circular Collider Design
Big Science Data


In Progress
Real-time cleansing of snow and weather data for operational avalanche forecasting
Energy, Climate & Environment


AI-augmented architectural design
Energy, Climate & Environment


In Progress
Extracting activity from large 4D whole-brain image datasets
Biomedical Data Science


Latest news

PassGPT | Using language models to enhance password security
February 6, 2024

PassGPT | Using language models to enhance password security

PassGPT | Using language models to enhance password security

PassGPT is a Large Language Model for password generation trained on leaked passwords, which can outperform existing methods based on generative adversarial networks by guessing twice as many unseen passwords.
ADORE | A benchmark dataset in ecotoxicology to foster the adoption of machine learning
January 24, 2024

ADORE | A benchmark dataset in ecotoxicology to foster the adoption of machine learning

ADORE | A benchmark dataset in ecotoxicology to foster the adoption of machine learning

Applying machine learning to ecotoxicology could help reduce the number of animal tests, costs, and animals sacrificed while preserving the accuracy of the in vivo tests.
License Flowers | Art and AI at SDSC
February 21, 2024

License Flowers | Art and AI at SDSC

License Flowers | Art and AI at SDSC

An adventure to create art using AI to raise awareness on code licenses

Contact us

Let’s talk Data Science

Do you need our services or expertise?
Contact us for your next Data Science project!