Mass spectrometry laboratories generate growing volumes of complex data, often across instruments from several vendors. Disconnected software, incompatible file formats, and difficult-to-access historical results can slow analysis and limit data reuse.
In this episode of Concentrating on Chromatography, produced in collaboration with Separation Science, host David Oliva speaks with Dr. Ansgar Korf, Chief Executive Officer of mzio GmbH, about how MZmine addresses these challenges through vendor-neutral processing, open-source development, AI, and automation.
From Open Source to Commercial Software
MZmine began as an academic project in 2004, when high-resolution mass spectrometry instruments started generating more data than existing software could manage. Its modular structure allowed researchers to add processing tools without altering the core platform, supporting an international open-source community.
Korf joined the project while completing his PhD in analytical chemistry. He needed stronger tools for lipid identification and chose to build on MZmine rather than create a separate platform. In 2023, the development team founded mzio to maintain the software, support its users, and guide its continued growth.
One Platform for Multivendor Data
Most modern laboratories use instruments from several manufacturers. They may also combine LC–MS, GC–MS, ion mobility, and other analytical techniques. Each system can require separate software, training, and data-processing workflows.
MZmine provides a vendor-neutral environment for processing these datasets. Its modular architecture allows developers to reuse core functions across techniques and add specialized tools where needed, such as spectral deconvolution for GC–MS.
This approach can reduce training demands and help laboratories select instruments based on analytical needs rather than software familiarity.
AI-Powered Data Reuse and Molecular Networking
The FAIR-MS initiative aims to make historical mass spectrometry data searchable and reusable. MZmine processes data from different instruments, while deep-learning models convert results into numerical representations called embeddings. Scientists can then compare spectra across platforms, projects, and time periods.
Deep learning also supports molecular networking. Traditional spectral matching can miss structurally related compounds when small chemical changes produce large differences in MS/MS spectra. Models such as DreaMS and MS2DeepScore can identify relationships that conventional similarity algorithms may overlook.
Korf stresses that AI should support scientific judgment rather than replace it. Scientists still need to inspect results, review the analytical evidence, and make final decisions.
Clearer Lipid and Small-Molecule Results
MZmine’s 2026 release includes dashboards for lipid and small-molecule analysis, as well as an impurity analysis workflow. The lipid dashboard combines MS1 data, MS/MS spectra, retention time, and related annotations to support more confident identification.
Users can review processed results and return to the underlying chromatograms and spectra without opening another software package. This connection to the raw data helps prevent automated analysis from becoming a black box.
Toward Automated MS Data Processing
Korf expects more data processing to occur on or near the acquisition instrument. Moving terabytes of raw data can take longer than the analysis itself, making data transfer a growing bottleneck.
Future workflows could optimize processing parameters, analyze data during acquisition, and provide scientists with faster feedback. Experts would retain control over key settings, while guided automation could make advanced mass spectrometry data analysis more accessible to new users.
The goal is a transparent workflow that reduces manual effort without removing scientists from the decision-making process.
Learn More:
- Explore the Concentrating On Chromatography podcast to dive into the frontiers of chromatography, mass spectrometry, and sample preparation with host David Oliva.
Connect with Ansgar:
- LinkedIn: Ansgar Korf
