33. Glossary¶

A reference for terms used throughout this specification.

Apache Arrow: An in-memory columnar data model that shares Parquet's columnar abstraction; Arrow and Parquet schemas map to one another and are straightforward to convert between.
Apache Parquet: The strongly-typed, compressed, columnar on-disk format in which each facet of an mzPeak archive is stored. See Anatomy of a Parquet file.
Archive Member: A file in an mzPeak archive, regardless of whether or not it is a Parquet file or not. An archive member SHOULD be listed in the index file
Array Index: The set of signal arrays mapped to a column or sub-tree. Encoded as a JSON object in the Parquet key-value metadata in data_arrays and peaks data_kinds for the relevant entity_type. The schema is defined in The Array Index.
Auxiliary Array: An array that did not run parallel to the primary axis of an entity or that simply was not mapped to a column. It is stored in the metadata table for the relevant entity_type.
Buffer Format: The disposition of the data in the column or sub-tree in a Signal Data layout file. See Buffer format.
Buffer Priority: Whether the column defines the primary representation of an array type or an alternative data type or unit. See Buffer priority.
Column Mapping: A data structure rule mapping a column in a metadata table to a controlled vocabulary term. Defined in Column mapping and stored in the index file.
Centroid: A spectrum (or peak list) reduced to discrete m/z–intensity peaks, as opposed to profile data.
Chromatogram: A measurement over time (for example a total ion current or extracted-ion trace); one of the mzPeak entity types. See Entity Type.
Chunked layout: A signal-data layout that cuts the sorted main axis into non-overlapping chunks (for example fixed m/z windows) — recording each chunk's start, end, and index — to allow random access along the axis. See Chunked Layout.
Controlled vocabulary (CV): A curated set of defined terms with stable accessions, used to annotate metadata unambiguously. mzPeak uses the PSI-MS controlled vocabulary.
CURIE: A compact URI of the form prefix:reference (for example MS:1000559) used to reference a controlled-vocabulary term.
Data kind: An index-file field declaring a member's semantics and expected schema family: data_arrays, peaks, metadata, proprietary, or other. See Data Kind.
Delta encoding: An opaque transform that stores successive differences instead of absolute values, compressing sorted axes such as m/z efficiently.
Entity type: An index-file field declaring what a member describes: spectrum, chromatogram, wavelength_spectrum, or other. See Entity Type.
Entity Index: The column in a data_arrays, peaks, or metadata table that is named <entity_type>_index. This column binds the row to a specific instance of the specified entity. These rows are usually the primary key or a component of a compound primary key, in relational database terms.
Facet: A file in an mzPeak archive defined by an entity_type and data_kind together.
Grid Encoding: A model that converts real-valued coordinate observations to a theoretical grid index and back again. These grid indices may come with some loss of precision, but can be much easier to compress. Defined in Grid encoding.
HUPO-PSI: The Human Proteome Organization Proteomics Standards Initiative, the body that governs mzPeak and maintains the PSI-MS controlled vocabulary.
imzML: The mzML-derived open standard for mass-spectrometry imaging data (XML plus an .ibd binary sidecar).
Index file (mzpeak_index.json): The JSON manifest listing every archive member with its entity_type and data_kind, so a reader resolves files by meaning rather than by name. See Index File.
Ion mobility: A gas-phase separation dimension (for example drift time or 1/K0) that may be stored as an array parallel to m/z and intensity.
m/z: Mass-to-charge ratio; the primary measurement axis of a mass spectrum.
mzML: The HUPO-PSI XML standard for mass-spectrometry data; mzPeak draws on its data model and reuses concepts such as controlled vocabularies where feasible.
mzPeak: An open, columnar mass-spectrometry data format: a ZIP archive (or directory) of Apache Parquet facets plus a JSON index, annotated using the PSI-MS controlled vocabulary under HUPO-PSI.
nativeID: A consistent, multi-component textual representation of an entity formatted to follow a vendor-specific pattern. These are governed by children of the native spectrum identifier format.
Null marking: A profile-data compression technique that replaces flanking zero-intensity points with null m/z and intensity values — so Parquet stores only a validity bit — reconstructing positions from a fitted m/z-spacing model. See Null Marking.
Numpress (MS-Numpress): A family of lossy and lossless numeric compression schemes for m/z and intensity arrays, usable in mzPeak as an opaque transform.
Opaque transform: A per-array encoding (for example delta encoding or Numpress) applied within a chunk and recorded so a reader can decode it. Some transforms (such as certain Numpress modes) are lossy.
Packed parallel metadata tables: The metadata layout that stores several related sub-tables (spectrum, scan, precursor, selected ion) side by side, linked by primary- and foreign-key index columns. See Packed Parallel Metadata Tables.
Page index: A Parquet footer structure — which mzPeak writers MUST emit — that enables random access below the row-group level.
Point layout: A signal-data layout that stores arrays as-is in parallel columns alongside a repeated entity-index column. See Point Layout.
Profile: A spectrum stored as a quasi-continuous trace of m/z–intensity samples, as opposed to centroid data.
PSI-MS CV: The controlled vocabulary of mass-spectrometry terms maintained by HUPO-PSI; the semantic backbone of mzPeak metadata.
Row group: A horizontal partition of a Parquet file — the unit of columnar compression and coarse random access.
Sorting rank: An array's ordering role within a signal layout; sorting rank 0 is the sorted "main axis" around which all parallel arrays are arranged. Arrays of differing length SHOULD instead be stored as auxiliary arrays.
Term marker: A column mapping that indicates the presence absence of a specific value-less term (boolean) OR a child of a specific term (string). Defined in Controlled Vocabulary Terms.
Total ion current (TIC): The summed intensity recorded for a spectrum; frequently traced over time as a chromatogram.
USI: The Universal Spectrum Identifier specification notation for referring to mass spectra in external collections. It is defined in https://www.psidev.info/usi.
Zero-run stripping: A profile-data size reduction that removes all but the first and last zero-intensity points in empty regions of a spectrum. See Zero Run Stripping.