Why Science? | mzML and Open Mass Spectrometry Data

Science learning becomes powerful when students can tell a measured pattern from a model, a statistical decision from a mechanism, and a useful lead from a finished conclusion. mzML is the Human Proteome Organisation Proteomics Standards Initiative open XML format for mass-spectrometry output and its acquisition and processing context. The official project explains that proprietary instrument formats and reduced peak lists can hinder comparison and omit precursor or instrument metadata. mzML represents spectra and chromatograms, can preserve profile or centroid data, uses controlled vocabulary terms for instruments and methods, and can encode binary numerical arrays with declared precision and compression. The current official page exposes a 1.1.1 main schema and a 1.1.3 optional index schema, while the full specification remains version 1.1.0. An open file improves interoperability; it does not restore information discarded during conversion or prove that an instrument run is scientifically valid. This guide supports Primary Science, PSLE Science, Secondary Science, O-Level Science and STEM education as a longform evidence-reading route; it is not medical advice, a laboratory protocol, an admission promise or a career guarantee.

Related eduKate reading: mass spectrometry, ionisation and mass-to-charge evidence; ion-mobility mass spectrometry; hydrogen–deuterium exchange mass spectrometry; Science Learning Hub; Education Hub; STEM Education; Careers by Subject Capability.

Primary and official sources checked on 12 October 2026: HUPO-PSI mzML official page and current schema resources, checked 12 October 2026; HUPO-PSI mzML official repository and specification, checked 12 October 2026; HUPO-PSI PSI-MS controlled vocabulary, checked 12 October 2026; 2026 Singapore–Cambridge O-Level Biology syllabus; MOE G2/G3 Lower Secondary Science teaching and learning syllabus; 2026 MOE G2 Computing syllabus. Specifications, vocabularies, tools and recommendations can change, so preserve dated documentation, exact source data and software versions.

A deeper reading route

Mass spectrometers produce signals under instrument-specific acquisition methods, often stored first in proprietary files. mzML provides a common exchange layer so different tools can inspect spectra, chromatograms and metadata without pretending every instrument works identically. The original vendor file, acquisition method and conversion receipt should remain available because the open representation is a translation, not the physical measurement itself.

An mzML document is XML with a defined schema and controlled-vocabulary annotations. Core lists describe instruments, software, samples, processing and run content. References connect repeated objects so a spectrum can point to the instrument configuration and data-processing history that apply. Structural validity confirms the document model; semantic validation checks that controlled terms appear in allowed contexts.

A mass spectrum normally contains paired arrays such as mass-to-charge values and intensities. The binary numbers are encoded inside XML, with metadata stating array type, precision, byte order and compression. A parser must decode every declaration correctly. Reading 64-bit values as 32-bit, treating an intensity array as an m/z array or applying compression twice can produce plausible-looking but wrong peaks.

Chromatograms use time-like coordinates and intensities to show signals across an acquisition. Total-ion, selected-ion and other chromatogram types answer different questions and are distinguished by controlled terms. Units matter: minutes and seconds can differ by sixty-fold while both produce smooth curves. Every numerical array needs its data type, unit and semantic role.

Profile and centroid spectra are not interchangeable. Profile data preserve sampled peak shapes, whereas centroiding summarises peaks into representative positions and intensities. HUPO-PSI notes that transformation from proprietary data to peak lists can include centroiding and deisotoping. Report whether conversion retained profile data, performed peak picking or used vendor algorithms.

MS level and precursor metadata connect fragmentation spectra to earlier ion selections. Isolation windows, selected-ion attributes, collision energy and activation method shape what fragments can appear. Missing or inaccurate precursor links can mislead identification software even when the spectrum arrays themselves decode correctly. Validate parent-child relationships and acquisition mode.

The PSI-MS controlled vocabulary supplies stable accessions for instruments, software, scan types, array meanings and many processing actions. Labels are useful to people, but the accession and vocabulary release provide machine identity. A locally invented phrase should use an appropriate user parameter only when no suitable controlled term exists and should never impersonate an official accession.

Data processing can contain several steps and software tools. Conversion, peak picking, charge deconvolution, smoothing and filtering may each alter the representation. The processing graph needs software name, version and parameters where available. A final mzML file should reveal whether a clean spectrum came from raw profile data or from aggressive preprocessing.

Indexed mzML adds an optional wrapper with offsets and a checksum so software can reach spectra without scanning the whole XML stream. The official page publishes a separate optional index schema. Index validity depends on the exact file bytes; editing or reformatting XML after index creation can invalidate offsets even when the inner mzML remains readable.

Validation should combine XML schema checks, controlled-vocabulary rules and numerical tests. Count spectra and chromatograms, verify array lengths, decode representative values, inspect index offsets, confirm units and compare selected spectra with the vendor file or an independent converter. Include empty arrays, high precision values, profile data, centroid data and multiple MS levels.

File size encourages compression and lower numerical precision, but each choice can affect downstream work. Zlib compression is lossless for the encoded bytes; changing 64-bit values to 32-bit reduces precision. Numpress or other specialised encodings need explicit terms and compatible readers. Benchmark size, speed and numerical differences rather than assuming smaller means equivalent.

Interoperability does not erase experimental context. Calibration, contamination, ion suppression, detector saturation and sampling design still affect spectra. mzML can carry metadata that helps evaluation, yet a complete file cannot turn a poorly controlled run into reliable biological evidence. Quality metrics and study design remain separate evidence layers.

For students, mzML is like a science notebook whose graphs, instrument settings and processing steps use a shared dictionary. Two laboratories can exchange the notebook because each label has an agreed meaning. The optimistic lesson is that open standards make evidence easier to inspect while still asking us to protect the original measurement.

A durable package contains vendor files or governed access, acquisition method, calibration and sample manifest, converter and version, command, mzML schema and CV releases, output hash, validation report and a few expected spectra. That package lets another analyst separate acquisition changes, conversion changes, XML changes and downstream analysis changes.

Versioning requires precision. The official mzML page currently exposes a 1.1.1 main schema and a 1.1.3 optional index schema, and retains the 1.1.0 specification document, while the old 1.0.0 material is explicitly marked obsolete. Cite the schema actually used by the file, the controlled-vocabulary release and the document consulted instead of reporting a vague mzML version.

Privacy and licensing can apply even to analytical files. Sample names, run paths and metadata may reveal participants, unpublished compounds or laboratory systems. De-identify only under a documented policy and avoid breaking provenance links. An open format does not automatically make a dataset appropriate for public release.

A strong mzML review separates what the format can preserve from what a laboratory actually preserved. The schema can describe spectra, chromatograms, arrays, instrument settings and processing, yet a converted file may still reflect earlier centroiding, peak filtering or metadata loss. Reviewers should therefore compare acquisition records, conversion logs and checksums with the final exchange file. That chain turns an apparently technical file-format choice into a practical lesson about evidence provenance, reproducibility and the limits of later reconstruction.

The controlled vocabulary also deserves active reading. A term accession is not decoration: it tells software and people what an array, unit, instrument component or processing step means. When a term, unit or precision is wrong, a file may remain syntactically readable while becoming scientifically misleading. Students can learn this distinction by comparing a file that passes an XML check with one whose scientific labels do not match the values. Structure and meaning are separate tests, and reliable data needs both.

For classroom use, a small fictional experiment is enough. Let one group document a spectrum, another a chromatogram and a third the conversion history, then ask a fourth group to reconstruct the evidence without verbal help. Missing scan context, undeclared units or ambiguous processing quickly becomes visible. The exercise connects chemistry and data analysis without pretending that reading XML replaces laboratory training. It also shows why open exchange standards help teams collaborate across instruments, software packages and institutions.

A careful publication should cite the exact mzML specification or schema version used and keep the original vendor data when policy and storage allow. Future software may interpret fields more accurately, but it cannot recover information that was never exported. This is the cheerful practical promise of good stewardship: preserving a clear source-to-conversion-to-analysis route gives tomorrow’s researchers more ways to ask new questions of today’s experiment.

A four-step claim ladder for this method

Stage	Question	Evidence	Responsible output
Acquire	What did the instrument record?	Spectra, chromatograms, scan and precursor metadata	A run-level measurement record
Encode	How are numeric arrays represented?	Array type, precision, units and compression	Decodable values
Describe	Which instrument and processing history apply?	Controlled terms, software, configurations and dataProcessing	Interpretable provenance
Judge	What can the file support?	Validation, controls, calibration and retained information	A bounded analytical claim