Written and maintained by CASRAI Editorial Board
Published · Last updated
On this page
OpenMS is an open-source, high-performance C++ software framework and algorithm library designed for the reproducible management, processing, and quantitative analysis of liquid chromatography-mass spectrometry (LC-MS) data across proteomics, metabolomics, and lipidomics research. Distributed under the permissive 3-clause BSD license, OpenMS provides researchers with more than 100 modular command-line executables known as The OpenMS Proteomics Pipeline (TOPP), direct Python scripting bindings via pyOpenMS, and seamless graphical pipeline construction through workflow engines such as KNIME and Nextflow. By natively supporting standard Human Proteome Organization Proteomics Standards Initiative (HUPO-PSI) open formats such as mzML, mzIdentML, and mzTab, OpenMS eliminates vendor data lock-in and scales from single-laboratory benchtop analyses to multi-terabyte cloud and high-performance computing pipelines.
What it is used for in research
Modern biomedical and biochemical investigations rely heavily on high-resolution mass spectrometry to profile complex biological matrices. Whether quantifying differential protein abundance across clinical cohorts or identifying altered metabolic pathways in disease models, raw spectral acquisitions require sophisticated signal processing before biological inferences can be drawn. Understanding the basic hardware configuration of a what is a mass spectrometer is essential, but the downstream computational pipeline dictates data fidelity, quantitative accuracy, and long-term analytical reproducibility.
OpenMS fulfills this computational requirement by serving as a modular, extensible backbone for liquid chromatography coupled to mass spectrometry (LC-MS) pipelines. In quantitative proteomics, OpenMS is deployed for both data-dependent acquisition (DDA) and data-independent acquisition (DIA) workflows. It enables automated peak picking, retention time alignment across multi-sample cohorts, peptide-spectrum match (PSM) scoring, false discovery rate (FDR) estimation using target-decoy databases, and label-free quantification (LFQ) as well as isobaric labeling (TMT and iTRAQ) quantification.
In metabolomics and untargeted small-molecule research, OpenMS addresses the challenges of chemical diversity and non-standard isotopic envelopes. Through specialized algorithms like FeatureFinderMetabo, the software detects low-molecular-weight metabolites, groups adducts, resolves co-eluting isobaric species, and performs multi-batch feature linking. Furthermore, OpenMS supports cross-omics and proteogenomics applications by integrating genomic sequence variants into custom identification databases and standardizing multi-omics outputs into community repository formats for submission to a designated domain repository.
How to install or access it
OpenMS is distributed across multiple channels depending on whether researchers require standalone command-line tools, visual workflow environments, or Python development libraries. Users should always obtain software packages directly from verified official repositories to ensure cryptographic integrity and access to current patch releases.
For computational biologists working in Linux, macOS, or Windows Subsystem for Linux (WSL2) environments, the recommended and most reproducible installation method is through Bioconda. Creating an isolated Conda environment prevents dependency conflicts with system-level C++ compilers or Python libraries:
- Configure Conda channels with proper priority:
conda config --add channels defaults conda config --add channels bioconda conda config --add channels conda-forge conda config --set channel_priority strict - Create a dedicated environment and install the core OpenMS package:
conda create -n openms-env python=3.11 openms conda activate openms-env - Verify the installation by querying a core TOPP utility:
FileInfo --help
For researchers developing custom algorithmic workflows in Python, the official pyOpenMS package is hosted directly on PyPI. It distributes pre-compiled binary wheels for x86_64 and ARM64 architectures across Linux, macOS, and Windows:
pip install numpy
pip install pyopenms
For bench scientists requiring graphical visualization tools, native desktop installers containing the complete TOPP executable suite, the TOPPView spectrum visualizer, and the TOPPAS graphical pipeline editor are provided on the official OpenMS project downloads portal (openms.de and github.com/OpenMS/OpenMS/releases). On macOS systems, Gatekeeper security attributes may quarantine downloaded application bundles; users can clear this attribute via the terminal command:
sudo xattr -r -d com.apple.quarantine /Applications/OpenMS-3.2.0
For large-scale enterprise or cluster execution, containerized Docker and Singularity images are maintained through the Biocontainers registry and the nf-core community (such as the nf-core/quantms workflow), guaranteeing identical runtime environments across heterogeneous high-performance computing clusters.
Step-by-step instructions
A foundational task in LC-MS untargeted analysis is label-free feature extraction and multi-run alignment. Below is a complete, reproducible five-step workflow demonstrating how to take centroided spectral data (mzML), detect chromatographic features, align retention times across analytical runs, link consensus features, and export tabular quantitative matrices.
Step 1: Convert and verify centroided spectral input
OpenMS algorithms require centroided mass spectra for accurate peak model fitting. If raw vendor files were converted in profile mode, execute PeakPickerHiRes to generate centroided data:
PeakPickerHiRes -in raw_profile_sample1.mzML -out sample1_centroided.mzML -algorithm:ms_levels 1,2
Inspect the file integrity using FileInfo to ensure spectral scan counts and retention time ranges are populated correctly:
FileInfo -in sample1_centroided.mzML -stats
Step 2: Detect chromatographic features with FeatureFinderCentroided
Group consecutive mass peaks across retention time into discrete two-dimensional chromatographic features using FeatureFinderCentroided. First, generate a default parameter template to inspect algorithm thresholds:
FeatureFinderCentroided -write_ini ff_defaults.ini
Execute feature detection across individual samples, allocating multi-threaded CPU execution:
FeatureFinderCentroided -in sample1_centroided.mzML -out sample1_features.featureXML -threads 8
FeatureFinderCentroided -in sample2_centroided.mzML -out sample2_features.featureXML -threads 8
For small-molecule metabolomics studies, substitute FeatureFinderMetabo, which uses isotope pattern fitting optimized for non-peptide molecular formulas.
Step 3: Align retention times with MapAlignerPoseidon
Chromatographic retention times naturally drift across continuous instrument runs. OpenMS provides several map alignment algorithms. Execute MapAlignerPoseidon to compute non-linear retention time warping functions across feature maps:
MapAlignerPoseidon -in sample1_features.featureXML sample2_features.featureXML -out sample1_aligned.featureXML sample2_aligned.featureXML -algorithm:max_difference 30.0
This step calculates a spline or polynomial transformation that maps elution times onto a common reference coordinate system without altering raw mass-to-charge ratios.
Step 4: Group features across samples with FeatureLinkerUnlabeledKD
Link corresponding features across the aligned multi-run cohort into a single consensus map using a KD-tree clustering algorithm:
FeatureLinkerUnlabeledKD -in sample1_aligned.featureXML sample2_aligned.featureXML -out consensus_runs.consensusXML -algorithm:mz_tol 10.0 -algorithm:rt_tol 20.0
The resulting consensusXML file contains aligned feature clusters, reporting matched retention times, mean m/z values, charge states, and individual sample intensities.
Step 5: Export quantitative tables for statistical analysis
Convert the XML consensus map into standard delimited text format for downstream differential expression analysis in R, Python, or statistical software packages:
TextExporter -in consensus_runs.consensusXML -out quantitative_matrix.csv -consensus:centroids -consensus:element:charge -consensus:element:rt
Alternative programmatic execution using pyOpenMS
The exact same feature extraction and file I/O operations can be executed interactively in Python using pyOpenMS. The following script demonstrates loading an mzML file, executing centroided feature finding, and querying extracted features:
import pyopenms as ms
# 1. Load centroided mzML experiment
exp = ms.MSExperiment()
ms.MzMLFile().load("sample1_centroided.mzML", exp)
print(f"Loaded {exp.getNrSpectra()} spectra from sample1")
# 2. Configure FeatureFinderCentroided
features = ms.FeatureMap()
ff = ms.FeatureFinder()
param = ms.Param()
ff.setLogType(ms.LogType.CMD)
# 3. Execute feature detection
seeds = ms.FeatureMap()
ff.run("centroided", exp, features, param, seeds)
print(f"Detected {features.size()} chromatographic features")
# 4. Save results to featureXML
ms.FeatureXMLFile().store("sample1_features.featureXML", features)
Output and how to read it
OpenMS produces both standardized community formats and native structured XML outputs. Understanding the structural hierarchy of these files allows researchers to diagnose data processing artifacts and verify quantitative accuracy.
The primary intermediate outputs generated during an OpenMS workflow include:
featureXML: Stores two-dimensional chromatographic features extracted from an individual LC-MS run. Each feature entry records a unique integer identifier, retention time centroid, monoisotopic m/z, overall intensity (calculated by integrating peak area), assigned charge state, and an algorithmic quality score ranging between 0 and 1. Subordinate elements document individual mass traces and convex hull boundaries defining the elution envelope.consensusXML: Contains groupings of corresponding features identified across multiple analytical runs. Each consensus feature element lists the centroid m/z and retention time across the cohort, followed by a map of run-specific feature references documenting sample-specific intensities and retention time shifts.idXML: Formats peptide-spectrum matches (PSMs) and protein inferences generated by integrated search engines (such as MS-GF+, Comet, Mascot, or X!Tandem). Each peptide identification record pairs an experimental MS2 spectrum index with candidate peptide sequences, calculated precursor mass deviations, search engine scores, and target-decoy false discovery rate (FDR) q-values.mzTaband CSV: Tabular summary formats designed for direct consumption by biostatistical packages. In an mzTab file, the small-molecule section (SML) or protein/peptide summary sections (PRT/PEP) provide structured columns containing normalized abundance values across experimental study variables.
To visually inspect outputs, launch TOPPView from the desktop terminal. Users can overlay a featureXML map directly on top of the original two-dimensional retention time versus m/z heat map from the mzML file. This visual inspection confirms whether chromatographic peak boundaries encompass the true isotopic distribution and ensures that tailing or split peaks were not erroneously partitioned into multiple spurious features.
Troubleshooting
Mass spectrometry workflows process massive data volumes and involve intricate multi-dimensional parameter spaces. Below are diagnostic strategies for addressing common technical failures encountered during OpenMS pipeline execution:
- Out-of-memory errors on large mzML files: High-resolution Orbitrap or timsTOF runs can produce mzML files exceeding 10 gigabytes. If a TOPP tool or pyOpenMS script terminates with a bad allocation error, ensure that spectra are strictly centroided rather than stored as dense profile points. When processing cohort-scale datasets, execute tools using
-threadstuned to available system RAM (typically allocating at least 4 gigabytes of memory per execution thread) or process runs using batch chunking in Nextflow. - Zero features detected by FeatureFinderCentroided: This issue typically occurs when input files contain profile data or when mass tolerance parameters are incompatible with the instrument resolution. Verify that precursor m/z tolerance (
algorithm:common:mass_error_ppm) matches instrument calibration: set to 5–10 ppm for Orbitrap or Q-TOF systems, but expand to 0.5–1.0 Da for low-resolution ion traps. EnsurePeakPickerHiReshas been executed if raw scans contain continuous profile envelopes. - Retention time alignment failure in MapAligner: If non-linear alignment produces distorted retention time warps or terminates with clustering errors, the maximum allowable retention time shift parameter (
algorithm:max_difference) may be too restrictive. In gradient runs subject to column aging or temperature fluctuations, inspect base peak chromatograms (BPCs) in TOPPView and expand the alignment search window to accommodate observed maximum retention shifts. - Conda library linkage errors on Linux: When running OpenMS installed via Bioconda on older Linux distributions, users may occasionally encounter
GLIBCXXor OpenMP linkage errors. Resolve this by forcing Conda to update runtime libraries:conda install -c conda-forge libstdcxx-ng libgompwithin the active environment. - macOS Gatekeeper application blocking: macOS security protocols may flag standalone binary executables or TOPPView as unsigned. In addition to running
xattr -r -d com.apple.quarantineon the installation folder, open System Settings, navigate to Privacy and Security, and approve the blocked binary under the Security header.
Free alternatives
While OpenMS provides an exceptionally comprehensive and modular C++ library suitable for automated high-throughput pipelines, researchers often evaluate several alternative open-source and free mass spectrometry data processing platforms. Depending on specific laboratory workflows, programming expertise, and targeted analytes, competing packages may offer specialized advantages.
For researchers immersed in statistical programming within the R ecosystem, the XCMS package in R is widely regarded as an industry standard for untargeted metabolomics feature detection, non-linear retention time alignment via obiwarp, and CAMERA-based isotopic grouping. However, XCMS lacks the extensive native proteomics search engine adapters and isobaric quantification modules present in OpenMS.
For laboratories prioritizing an intuitive graphical user interface for metabolomics and lipidomics, the MZmine platform offers a flexible, modular Java architecture with interactive chromatogram inspection, modular batch queuing, and native support for ion mobility mass spectrometry (IM-MS) multidimensional data.
In high-throughput proteomics, MaxQuant represents a ubiquitous freeware solution providing integrated Andromeda database searching and MaxLFQ quantification algorithms, although its closed-source core prevents direct programmatic inspection and custom C++ extension. For ultrafast database matching, the FragPipe proteomics suite pairs MSFragger with Philosopher and IonQuant, achieving rapid search speeds on large-scale cohort datasets. Meanwhile, Skyline provides an open-source Windows-centric desktop environment tailored for targeted multiple reaction monitoring (MRM) and parallel reaction monitoring (PRM) assay development.
The data table below compares the architectural specifications, licensing models, CLI capabilities, and primary analytical applications across these leading mass spectrometry processing platforms.
Frequently asked questions
What is the difference between OpenMS and pyOpenMS?
OpenMS is the core C++ framework containing underlying data structures, file parsers, and algorithmic libraries, along with the command-line TOPP executables. In contrast, pyOpenMS provides Python bindings generated via Cython that expose OpenMS C++ classes directly to Python, allowing researchers to write lightweight scripts and integrate mass spectrometry algorithms directly with NumPy, SciPy, and Pandas.
Does OpenMS support raw proprietary vendor files?
OpenMS primarily operates on open standard formats, particularly mzML. To process proprietary vendor files from Thermo Fisher, Bruker, Sciex, Waters, or Agilent, data must first be converted to centroided mzML using ProteoWizard msconvert. In Windows desktop environments where vendor dynamic link libraries (DLLs) are installed, OpenMS can invoke msconvert transparently in the background.
Can OpenMS run in cloud and high-performance computing (HPC) environments?
Yes. Because every TOPP tool is packaged as an independent, non-interactive command-line executable with standardized input/output flags, OpenMS is optimized for HPC cluster scheduling (such as Slurm and SGE) and cloud workflow engines. Containerized Nextflow pipelines like nf-core/quantms deploy OpenMS across distributed nodes without requiring local graphical dependencies.
What is the difference between featureXML and consensusXML?
A featureXML file stores two-dimensional chromatographic features (m/z, retention time, charge, and integrated abundance) detected within a single LC-MS run. A consensusXML file represents the output of multi-run feature linking, grouping corresponding features across multiple samples into a unified matrix to facilitate cohort-wide comparative quantification.
How does OpenMS handle decoy database generation for false discovery rate control?
OpenMS includes the DecoyDatabase utility, which accepts a target protein FASTA file and generates randomized, reversed, or shuffled decoy sequences while preserving peptide cleavage sites and amino acid composition. Downstream tools like FalseDiscoveryRate and IDFilter use these target-decoy distributions to calculate empirical q-values and filter peptide-spectrum matches at designated FDR thresholds.
How do I resolve memory exhaustion when processing large cohort datasets in OpenMS?
Memory exhaustion can be prevented by ensuring raw spectral data is centroided using PeakPickerHiRes before feature detection, setting conservative thread counts to ensure each process receives 4 to 8 gigabytes of RAM, and using disk-backed caching options in pyOpenMS when streaming high-resolution multi-file experiments.
Sources
- Pfeuffer, J., Bielow, C., Wein, S., et al. (2024). OpenMS 3 enables reproducible analysis of large-scale mass spectrometry data. Nature Methods, 21(3), 365–367. https://doi.org/10.1038/s41592-024-02197-7 (Checked 2026-11-15).
- Röst, H. L., Sachsenberg, T., Aiche, S., et al. (2016). OpenMS: a flexible open-source software platform for mass spectrometry data analysis. Nature Methods, 13(9), 741–748. https://doi.org/10.1038/nmeth.3959 (Checked 2026-11-15).
- Röst, H. L., et al. (2014). pyOpenMS: an open-source Python library for mass spectrometry. Bioinformatics, 30(17), 2540–2542. https://doi.org/10.1093/bioinformatics/btu394 (Checked 2026-11-15).
- OpenMS Documentation and User Guide (v3.x). OpenMS Team. https://openms.readthedocs.io/en/latest/ (Checked 2026-11-15).
- pyOpenMS Documentation (v3.x). OpenMS Team. https://pyopenms.readthedocs.io/en/latest/ (Checked 2026-11-15).
- OpenMS GitHub Repository. OpenMS Consortium. https://github.com/OpenMS/OpenMS (Checked 2026-11-15).
- Martens, L., et al. (2011). mzML—a community standard for mass spectrometry data. Molecular and Cellular Proteomics, 10(1), R110.000133. https://doi.org/10.1074/mcp.R110.000133 (Checked 2026-11-15).
Data: Open-Source Mass Spectrometry Data Processing Software Comparison
Comparative specifications, algorithmic capabilities, licensing, supported operating systems, and data formats across prominent open-source and free mass spectrometry data analysis platforms, compiled from official software documentation and peer-reviewed release papers.
Period covered: 2024/2026. Coverage: Global.
| Platform | Primary Focus | Software License | Core Language | CLI Automation | Scripting API | GUI Environment | Primary Input Format |
|---|---|---|---|---|---|---|---|
| OpenMS | Proteomics and Metabolomics | 3-clause BSD | C++ | Yes (TOPP tools) | Python (pyOpenMS) | Yes (TOPPView / TOPPAS) | mzML, mzXML |
| XCMS | Metabolomics | GPL-2.0 or later | R / C++ | Yes (R scripts / CLI) | R (xcms package) | Web (XCMS Online) | mzML, mzXML, netCDF |
| MZmine 3 | Metabolomics and Lipidomics | GPL-2.0 | Java | Yes (Batch CLI mode) | None (Java API) | Yes (Native Java GUI) | mzML, mzXML, netCDF, IM-MS |
| MaxQuant | Proteomics | Freeware (Proprietary) | C# (.NET) | Yes (MaxQuantCmd.exe) | None (R post-processing) | Yes (Native Windows GUI) | Thermo RAW, mzML, Bruker .d |
| FragPipe | Proteomics | Academic Free / GPL components | Java / Python | Yes (CLI runner) | Python / R interfaces | Yes (Java GUI) | mzML, Thermo RAW, Bruker .d |
| Skyline | Targeted Proteomics and Metabolomics | Apache 2.0 | C# (.NET) | Yes (SkylineCmd.exe) | Python / R tool store | Yes (Native Windows GUI) | Vendor RAW, mzML, mzXML |
| MS-DIAL | Metabolomics and Lipidomics | BSD-2-Clause | C# (.NET) | Yes (CLI version) | None (R / Python scripts) | Yes (Native Windows GUI) | mzML, AB Sciex, Waters, Agilent |
Sources
- OpenMS 3 enables reproducible analysis of large-scale mass spectrometry data (Nature Methods 2024) (accessed 2026-11-15)
- OpenMS Official User Guide and Architecture Documentation (accessed 2026-11-15)
- XCMS: LC-MS Data Processing in R (accessed 2026-11-15)
- MZmine 3 User Manual and Architecture Overview (accessed 2026-11-15)
- FragPipe Proteomics Pipeline Documentation (accessed 2026-11-15)
- Skyline: an open source document editor for targeted proteomics (accessed 2026-11-15)
Compiled by CASRAI from the sources above. Free to reuse under CC BY 4.0 with attribution.







