File Formats

OpenMS reads spectra from mzML and stores its results in its own XML formats: featureXML for features, consensusXML for features linked across runs, and idXML for identifications. This page describes three additions of OpenMS 3.6: compressed input and output, Parquet versions of the result formats, and imaging data. For Thermo .raw files and Bruker .d directories, see Vendor formats.

Compressed files

Reading

Tools read compressed XML files directly, without unpacking them first. This works for mzML, mzXML, mzData, featureXML, consensusXML, idXML, pepXML, protXML, traML, trafoXML and INI files, among others. Add .gz (gzip), .bz2 (bzip2) or .zip to the file name, for example sample.mzML.gz. A ZIP archive must contain exactly one file.

Note

In OpenMS 3.6, mzIdentML files (.mzid) cannot be read compressed. Unpack them first.

Other formats, for example MGF or text files, have to be unpacked before a tool can read them; a tool refuses a compressed file of such a format. A Bruker .d directory packed into a ZIP archive (.d.zip) is described in Vendor formats.

Writing

mzML, mzXML, mzData, featureXML, consensusXML, traML and mzIdentML output is compressed when the output file name ends in .gz (gzip) or .bz2 (bzip2):

FileConverter -in sample.mzML -out sample.mzML.gz

The other formats, idXML and trafoXML among them, cannot be written compressed, and ZIP output is not supported. A tool refuses such an output name with an error rather than write an uncompressed file under it: give the output an uncompressed name, and compress the file afterwards if needed. The same applies to the modes that write spectra one by one, for example FileConverter with -process_lowmemory: they write uncompressed mzML only.

Compressing the peak data inside mzML

The peak data inside an mzML file can be compressed as well. With -zstd_compression, new in OpenMS 3.6, FileConverter compresses it with Zstandard (zstd); FileFilter has the same option as -peak_options:zstd_compression true:

FileConverter -in sample.mzML -out sample_zstd.mzML -zstd_compression

zstd is lossless and usually gives smaller files than zlib, and it can be combined with the lossy numpress compression of -lossy_compression. Tools read such files like any other mzML file. Not every program outside OpenMS reads zstd-compressed mzML yet, so leave the option off for files that other software has to read.

Parquet bundles

OpenMS 3.6 can store identifications, feature maps and consensus maps as Apache Parquet tables instead of idXML, featureXML and consensusXML. Such a bundle is a directory of Parquet files:

Format	Instead of	Files in the directory
.idparquet	idXML	psms.parquet, proteins.parquet, protein_groups.parquet, search_params.parquet
.featureparquet	featureXML	features.parquet and the four files of .idparquet
.consensusparquet	consensusXML	consensus_features.parquet and the four files of .idparquet

Any Parquet reader can load these tables without OpenMS, for example pandas or pyarrow in Python, the arrow package in R, or DuckDB.

These tools read or write the bundles:

• .idparquet: CometAdapter, FeatureFinderIdentification, IDFileConverter, IDFilter, IDMerger, IDRipper, IDScoreSwitcher, IsobaricWorkflow, MapAlignerIdentification, MapRTTransformer, MS1LabeledWorkflow, MSGFPlusAdapter, MzTabExporter, PeptideIndexer, PercolatorAdapter, PSMFeatureExtractor, SageAdapter and TextExporter.

• .featureparquet: FeatureFinderIdentification, the FeatureLinker tools, FileConverter, IDConflictResolver, MapAlignerIdentification, MapRTTransformer, MzTabExporter and TextExporter.

• .consensusparquet: the FeatureLinker tools, FileConverter, IDConflictResolver, IDFilter, MapAlignerIdentification, MapRTTransformer, MzTabExporter and TextExporter.

To convert between the XML formats and the bundles, use FileConverter or ParquetConverter for feature and consensus maps, and IDFileConverter for identifications:

FileConverter -in sample.featureXML -out sample.featureparquet
IDFileConverter -in sample.idXML -out sample.idparquet

ParquetDiff compares two Parquet files, for example the same table of two bundles. It matches rows by a key and compares values with a tolerance, so a different row order or tiny floating-point differences do not count as changes.

Imaging data (imzML)

OpenMS 3.6 reads and writes imzML, the standard format for mass spectrometry imaging, in its library and in pyOpenMS. The TOPP tools and TOPPView do not accept imzML files yet.