eMZed 3: flexible and interactive development of scalable LC-MS/MS data analysis workflows in python Authors: Uwe Schmitt, Jethro Hemmann, Nicola Zamboni, Julia A Vorholt, Patrick Kiefer, Guoqiang Yu Published: 2026 ABSTRACT AbstractSummaryLiquid chromatography–mass spectrometry (LC-MS/MS) data analysis requires adaptable software solutions to meet diverse analytical needs. We present eMZed 3, a modern Python framework for flexible and interactive analysis of LC-MS/MS data. eMZed 3 enables users to develop scalable workflows tailored to their specific requirements while leveraging Python’s extensive ecosystem of libraries. Building on its predecessor, eMZed 3 is now Python 3-based and includes substantial enhancements, including support for chromatogram-based LC-MS data, a new SQLite-based backend supporting optional out-of-memory processing, and rich interactive visualization tools. Compared to the previous version, eMZed 3 is now split into three packages: emzed (core functionalities), emzed-gui (interactive data visualization), and emzed-spyder (an integrated development environment). This modular architecture allows straightforward integration of the emzed core library into headless Python environments, including computational notebooks (such as Jupyter) or high-performance computing clusters. eMZed 3 incorporates well-established libraries such as OpenMS, and is suited for both targeted and untargeted metabolomics. Overall, eMZed 3 supports the efficient development of scalable and reproducible LC-MS data analysis and is accessible to both novice and advanced programmers.Availability and implementationeMZed 3 and its documentation are freely available at https://emzed.ethz.ch, the source code is hosted at https://gitlab.com/groups/emzed3.An online-executable example workflow is available on Binder at: https://mybinder.org/v2/gl/emzed3%2Femzed-example-workflow/HEAD?_labpath=example.ipynb. 1 Introduction Liquid chromatography–mass spectrometry (LC-MS) is a core analytical technique in many fields of biology, environmental sciences, and chemistry. Advances in LC-MS instrumentation have enabled the generation of increasingly large and complex data sets, leading to computational challenges. Consequently, reliable and reproducible analysis of LC-MS data has become a critical step in the analytical workflow and is supported by a variety of commercial and open-source software tools. These tools often belong to one of two extremes: user-friendly software with limited flexibility, and libraries that require advanced programming skills but offer greater versatility. For example, MZmine 3 (Schmid et al. 2023) is a widely-used GUI-based application allowing the simple setup of data analysis workflows consisting of pre-defined processing steps. However, it offers limited flexibility and lacks easy programmatic customization. In contrast, the OpenMS project (Pfeuffer et al. 2024) provides a robust and performant C++ library with Python bindings that allows high flexibility in workflow programming, but lacks advanced interactive visualization capabilities. To bridge this gap and support the interactive prototyping and development of tailored workflows, we had introduced eMZed as a novel open-source Python framework (Kiefer et al. 2013). Since then, eMZed has been used within a number of research projects (Hartl et al. 2020, Martano et al. 2020, Nguyen et al. 2020, Reiter et al. 2020, Hemmann et al. 2021, Maier et al. 2021, Butin et al. 2022, Keller et al. 2022, Ryback et al. 2022, Reiter et al. 2024, Mori et al. 2025) and continued to be further developed. Here, we present a completely revised and modernized version 3 of eMZed. eMZed 3 has been rewritten for compatibility with Python 3 (Van Rossum and Drake 2009) and features numerous new functions and performance enhancements, including support for chromatogram-based LC-MS/MS data (e.g. multiple reaction monitoring data), out-of-memory processing, and improved interactive data inspection and visualization. The framework retains its core strength of offering a comprehensive set of elementary functions and data types for reading, processing, and analyzing LC-MS data. With an intuitive Python API, eMZed 3 allows users to create custom workflows that seamlessly integrate with popular Python libraries, such as pandas (McKinney 2010), NumPy (Harris et al. 2020), or scikit-learn (Pedregosa et al. 2011). It also enables interactive visualization of chromatograms, spectra, and peakmaps at any stage of a workflow. The optional Spyder-based integrated development environment (IDE) bundles the eMZed modules, helping programming beginners to quickly develop their own LC-MS data analysis workflows. eMZed supports all aspects of reproducible and tailored data analysis for both targeted and untargeted metabolomics, as well as other applications. 2 Results 2.1 Functionalities eMZed was designed as an easy-to-use coding framework that allows users to develop highly customizable workflows for LC-MS data analysis in Python (Fig. 1). To that end, eMZed builds upon existing algorithms from e.g. PyOpenMS (Röst et al. 2014) while incorporating additional features, including a versatile table data structure to store and access LC-MS peak lists and GUI-tools for the interactive inspection and re-integration of peaks. While eMZed includes many features found in widely used tools such as MZmine (Schmid et al. 2023) and XCMS (Louail et al. 2025) strength lies in enabling users to program custom workflows beyond the scope of existing tools and to address specific problems requiring tailored solutions (see also example workflow below). The core concept of eMZed. eMZed is designed as a modular framework (emzed, emzed-gui, emzed-spyder) to enable the Python-based development of highly customizable processing workflows for LC-MS/MS data. It incorporates key functions to access and process peaks, spectra, and chromatograms and connects with commonly used libraries from the Python ecosystem. The new version 3 of eMZed builds on the architecture and core features of the previous version of eMZed (Kiefer et al. 2013), but is now compatible with Python 3 and fully separated into three individual packages (all available at PyPI): emzed, emzed-gui, and emzed-spyder. The package emzed provides core functionalities of eMZed, while the optional emzed-gui allows interactive data inspection and visualization (see Fig. 2A for example code). To facilitate the interactive development of workflows with eMZed, emzed-spyder bundles a customized version of the Spyder IDE with the eMZed packages. For Windows, an installer is available at the eMZed website that allows an easy installation of eMZed 3 (emzed-spyder with all other eMZed packages). Visualization of LC-MS data using eMZed 3. A, Example code demonstrating a typical workflow: loading of LC-MS data, inspection of the peakmap, extraction and integration of LC-MS peaks corresponding to specific metabolites, and peak visualization. B, Screenshot of Peakmap Explorer, which allows the interactive inspection of LC-MS/MS peakmaps. C, Screenshot of Table Explorer, used to interactively inspect and (re-)integrate LC-MS peaks which are defined in an eMZed Table by their m/z and retention time values. The Python package emzed offers the core functionalities of eMZed 3, and does not depend on any GUI components. Thus, emzed can be used with any type of Python installation, including high-performance computing clusters or Jupyter notebooks (Kluyver et al. 2016). Among many features, emzed provides an easy-to-use API for: Direct access to chromatograms, peaks, and spectra (MSn) and their underlying data Direct access to chromatograms, peaks, and spectra (MSn) and their underlying data Targeted peak extraction Targeted peak extraction Untargeted feature finding, isotope, and adduct grouping Untargeted feature finding, isotope, and adduct grouping Alignment of peaks (m/z and retention time) and feature lists Alignment of peaks (m/z and retention time) and feature lists Powerful table-based operations on peak lists Powerful table-based operations on peak lists Handling of chemical data such as molecular formulas Handling of chemical data such as molecular formulas Accessing R functionalities using integrated bridge to R Accessing R functionalities using integrated bridge to R Several integration algorithms are provided by emzed and allow fast, parallel integration of chromatographic peaks. For untargeted analysis, eMZed integrates the feature finding algorithms FeatureFinderMetabo from OpenMS (Röst et al. 2016) and centWave (Tautenhahn et al. 2008), which is accessible from emzed via a builtin bridge to R (R Core Team 2021). Additionally, the ADAP feature finder from MZmine 2 (Myers et al. 2017) is available as an eMZed extension. eMZed 3 now also supports chromatogram-based LC-MS data, e.g. MRM data from triple quadrupole instruments. The powerful Table data structure plays a key role in eMZed, allowing operations such as filtering, grouping, joining, and aggregating and it now uses an SQLite database (Gaffney et al. 2022) as backend for improved performance and optional out-of-memory operations for processing large data sets. For interactive visual inspection of emzed data structures, the emzed-gui package provides several tools that have been improved in eMZed 3. The Peakmap Explorer allows quick exploration of entire LC-MS/MS peakmaps using a 2D heatmap for visualization and can display chromatograms as well as mass spectra (Fig. 2B). The Table Explorer is a central element of eMZed and assists in inspecting and modifying data stored in emzed tables (Fig. 2C). A prominent use case for a Table is the management of peak lists, and in this case the Table Explorer provides interactive visualizations and modifications of extracted ion chromatograms and mass spectra (e.g. for manual re-integration of peaks). In eMZed 3, Table Explorer now offers improved speed and robustness and more features, e.g. the coloring of individual rows or cells. emzed-gui also comes with tools to create simple graphical input forms that can be used in a workflow for user input. The optional emzed-spyder package is a customized IDE to facilitate interactive workflow development and prototyping. emzed-spyder is based on Spyder 5 and integrates all other eMZed packages. It supports direct inspection of PeakMap and Table data structures in the variable explorer and includes functionalities for the development of eMZed-based Python projects (also called eMZed extensions) based on separate virtual environments created for that purpose. 2.2 Implementation details For the development of eMZed 3, the previous Python 2-based code (Kiefer et al. 2013) was revised and modernized to be compatible with Python 3 (tested with versions 3.11–3.14). While most of the architecture from the previous version of eMZed was kept, the new implementation includes several notable technical improvements. To store and process information within the eMZed data structures Table and PeakMap, an SQLite database is now used internally, which results in a major performance enhancement and significantly reduced memory footprint. The SQLite database is per default kept in memory but optionally allows out-of-memory operations on disk, thus circumventing working memory limitations that might occur for large data. To allow access to functionalities from OpenMS (e.g. I/O operations, feature finding, spectral processing, retention time alignment), PyOpenMS (version 3.3.0) is used, which is run in its own process to prevent version conflicts between shared libraries, such as the Qt framework. The ADAP peak picker from MZmine 2 (Myers et al. 2017) is offered as a separate extension module. The GUI tools of eMZed are now migrated to PyQt5. The customized IDE emzed-spyder is based on Spyder (version 5.5.6) and enhances Spyder to support eMZed-specific features and offers functions for the development of eMZed-based projects. eMZed 3 runs on all modern operating systems (tested on Windows, Ubuntu, macOS), and the core library emzed is compatible with (cloud-based) notebook environments (e.g. Jupyter, Google Colab). 2.3 Example workflow To assist new users, we provide a comprehensive tutorial for beginners on the eMZed website (https://emzed.ethz.ch). In addition, we created a more advanced example workflow to demonstrate the capabilities of eMZed 3. The example workflow can be executed as a cloud-based notebook on Binder1 and highlights how eMZed can be used to correct for retention time (RT) shifts using a problem-specific approach. In this example, we consider the analysis of amino acids in a previously published data set comprising 200 samples (Maier et al. 2021). Raw data visualization reveals RT instability within a smaller sample subset. However, simply enlarging the RT windows used for peak integration is not a suitable approach to cope with the observed RT shifts, since the RT difference of the two mass isomers leucine and isoleucine is smaller than the observed maximal inter-sample RT shift. The example workflow thus implements a simple, case-specific solution that fixes the issue by using RT shifts of other amino acids as estimates to shift the RT windows for leucine and isoleucine. In the example notebook, we further evaluate whether the strategy was successful and visualize treatment-specific differences of the amino acid profiles. 3 Discussion eMZed seeks to fill a gap between other commonly used open-source tools for LC-MS data analysis, such as MZmine (Heuckeroth et al. 2024), OpenMS (Pfeuffer et al. 2024), MetaboAnalyst (Pang et al. 2024), or XCMS (Louail et al. 2025) by providing a powerful set of low-level functions to access, process, and interactively visualize LC-MS data that seamlessly integrates into the Python ecosystem of libraries for data analysis, machine learning, and plotting. By combining these functionalities, users can quickly develop tailored LC-MS data analysis workflows, while staying close to the raw data (chromatograms and spectra) with the help of the interactive GUI tools. We envision that eMZed 3 will be of particular benefit to users who require customized, yet robust and scalable data analysis pipelines, e.g. for analytical core facilities. We aim to continuously develop eMZed and provide additional eMZed-based tools and workflows in the future. As an open-source software, we particularly welcome community contributions (in the Gitlab repository). Supplementary Material Acknowledgements We acknowledge Andrea Zamuner for testing eMZed 3 and providing input for improvements. Footnotes https://mybinder.org/v2/gl/emzed3%2Femzed-example-workflow/HEAD?labpath=example.ipynb Contributor Information Uwe Schmitt, Scientific IT Services, ETH Zurich, Binzmuehlestrasse 130, Zurich 8092, Switzerland. Jethro Hemmann, Junior Research Group Metabolomics-Guided Natural Product Discovery, Leibniz Institute for Natural Product Research and Infection Biology (Leibniz-HKI), Beutenbergstraße 11a, Jena 07745, Germany. Nicola Zamboni, Institute of Molecular Systems Biology, ETH Zurich, Otto-Stern-Weg 3, Zurich 8093, Switzerland. Julia A Vorholt, Institute of Microbiology, ETH Zurich, Zurich, Vladimir-Prelog-Weg 4, Zurich 8093, Switzerland. Patrick Kiefer, Institute of Microbiology, ETH Zurich, Zurich, Vladimir-Prelog-Weg 4, Zurich 8093, Switzerland. Supplementary material Supplementary data are available at Bioinformatics Advances online. Conflicts of interest None declared. Funding This work was supported by ETH Zurich, Department of Biology, within the frame of an IT-strategy initiative; the Personalized Health and Related Technologies (PHRT) initiative of the ETH Domain; and the NCCR Microbiomes by the Swiss National Science Foundation [51NF40_180575, 51NF40_225148]. J.L.H. was supported by the Free State of Thuringia and the European Social Fund Plus [2024 FGR 0017]. Data availability statement No new data were generated or analysed in support of this research. References Associated Data Supplementary Materials Data Availability Statement No new data were generated or analysed in support of this research.