Skip to content

mscompress

Multi-threaded lossless and lossy compression for Mass Spectrometry data.

mscompress defines the .msz file format — a random-access compressed container for mzML data — and ships the C library, CLI, Python bindings, and Node.js/TypeScript bindings that read and write it. An extended container format, .mszx, bundles a compressed spectra file with search-result annotations and a JSON manifest.

What it does

  • Reads mzML (uncompressed XML) and produces MSZ (compressed, random access without full decompression).
  • Stores XML metadata, m/z arrays, and intensity arrays as three independently compressed streams so any spectrum can be retrieved in O(1) seeks.
  • Supports a registry of pre-compression transforms (delta encoding, log, bit-packing, type casts, variable bit-rate) that improve compression ratio on numeric arrays.
  • Bundles MSZX archives that ship search results (pepXML / Percolator TSV) alongside the spectra.

Pick your interface

  • Python

    Cython-backed bindings with optional Parquet, PyTorch, and JAX adapters.

    Get started

  • Node.js / TypeScript

    N-API addon with prebuilt binaries for macOS, Linux, and Windows.

    Get started

  • CLI

    Standalone mscompress binary for batch compression and extraction.

    Get started

  • C library

    Core library and architecture — how the format and algorithms work.

    Architecture