Research

From raw spectra to confident, comparable molecules

We develop efficient algorithms, tools and workflows for high-throughput mass spectrometry of small molecules. Our strength is fast, exact identification; we apply it to real questions in human, microbial and plant lipidomics, and increasingly complement it with machine learning and cloud-native analysis.

01

Harmonized naming and representation of ambiguity

AlgorithmsNomenclatureAnnotation

Fast, exact algorithms for annotating small molecules are the core of our work. We treat lipid names and structures as formal objects that software can parse, normalize and compare — so identifications are consistent across labs, tools and databases.

Grammar-based parsing and normalization of lipid shorthand nomenclature (Goslin).
Resolving isobaric and structurally similar compounds that confound naïve matching.
Shared libraries (C++, Python, Java, R) so every tool identifies the same way.
02

Targeted & quantitative workflows

Targeted MSQuantificationAssay design

Alongside untargeted discovery, we build reliable targeted and quantitative methods — the assays and spectral libraries needed to measure defined panels of lipids reproducibly.

Designing transition lists and in-silico spectral libraries (LipidCreator).
Stable-isotope labelling and collision-energy optimization for robust quantification.
Integration with established pipelines such as Skyline.
03

Comparison, ML & cloud analysis

Lipidome comparisonMachine learningCloud-native

We turn results into interpretable comparisons between whole lipidomes, and are beginning to complement our exact algorithms with machine learning, delivered through reproducible, cloud-native workflows.

Structural-space models that quantify how similar two lipidomes are (LipidSpace).
Feature selection to find the lipids that explain a study variable.
Containerized pipelines and browser-based dashboards for analysis at scale.
04

FAIR data & standards

mzTab-MmzQCHUPO-PSI

Methods only matter if their results can be shared and reused. We help define the community data standards that make small-molecule mass spectrometry interoperable and AI-ready.

Co-developing reporting formats for quantitative results (mzTab-M).
Quality-control metrics for MS runs and datasets (mzQC).
Contributing to HUPO-PSI work on AI-readiness and exchange formats.
Where we apply it

One toolset, many lipidomes

Human health

Plasma and tissue lipidomics to study disease mechanisms, biomarkers and lipid signalling.

Microbial

Characterising microbial lipidomes and their roles in metabolism and adaptation.

Plant

Plant lipidomics to connect lipid composition with physiology and environmental response.

Tools
The software we build →
Standards
Making MS data FAIR →
Publications
Ideas & findings →