Industry Roundup: mzmine's mzassist Brings AI to Mass Spectrometry Workflows

Bremen, Germany-based mzio GmbH issued a press release¹ announcing the launch of mzassist, an artificial intelligence (AI) and automation layer that lets mass spectrometrists run mzmine processing workflows through an AI assistant of their choosing. It ships with mzmine 4.11 this month.

mzassist is built on the model context protocol (MCP), an open standard Anthropic introduced in November 2024 for connecting AI applications to external tools. Through an MCP server, any compatible assistant can drive mzmine: importing raw data, selecting templates, configuring parameters, running batches and summarizing results. Users can hand over a whole workflow or step in manually at any stage.

The tool covers liquid chromatography–mass spectrometry (LC-MS), gas chromatography (GC) –MS and matrix-assisted laser desorption/ionization (MALDI) –MS data. Researchers got early hands-on access at a workshop during the 8th Munich Metabolomics Meeting on October 13^th of this year. Beyond running workflows, the assistant can explain processing steps, algorithms and parameters, and support spectral library building, which mzio says will help labs get new and existing methods running faster.

The Model Doesn't Make the Scientific Calls

The design addresses two worries labs have about AI: data leaving the building and results that can't be reproduced. Raw spectra stay on the lab's own computer, and users choose what information the assistant sees through opt-in and opt-out controls. The AI is active only when a user switches it on.

Crucially, parameter decisions are made by mzmine's deterministic algorithms, not the language model. Every analysis started through mzassist is saved as a complete mzmine workflow file recording each step and parameter, so a run can be audited and repeated without the AI.

Labs also connect whichever assistant their IT and data protection teams have already approved, rather than adopting a new one. "The lab chooses which AI assistant to trust, and mzmine stays the authoritative environment for the data analysis itself," said Ansgar Korf, CEO and co-founder of mzio.

Data-driven Parameter Setting

Two new tools target one of the most time-consuming parts of untargeted processing: choosing parameters. A parameter estimator analyzes a representative selection of a laboratory's raw files and derives processing settings from the signals themselves. An auto optimizer then benchmarks alternative settings across samples and identifies which perform best on the actual data.

Both tools can be used directly in mzmine, without any AI, or called by a connected assistant. "Parameter estimation and optimization in mzassist are data-driven algorithms, not guesswork by a language model," said Steffen Heuckeroth, CTO and co-founder of mzio, adding that the estimator gives consistent results every time.

Availability

mzassist is included in mzmine 4.11, released in October 2026. Academic users get it in mzmine COMMUNITY; commercial customers get it in mzmine PRO. mzmine, in development since 2005, is an open-source, vendor-neutral platform for processing LC-MS, GC-MS, ion mobility and imaging MS data, widely used in metabolomics, lipidomics and natural products research. More details are at mzio.io.

References

1. mzio Launches First Vendor-neutral AI Control Layer for Multimodal Mass Spectrometry Data Processing. mzio press release. (accessed 2027-10-27).