datarepo v1.3.2

Self-contained C# executables, built and verified by the dotnet-binaries workflow. See CHANGELOG.md.

datarepo v1.3.1

Self-contained C# executables, built and verified by the dotnet-binaries workflow. See CHANGELOG.md.

datarepo v1.3.0

Self-contained C# executables, built and verified by the dotnet-binaries workflow. See CHANGELOG.md.

datarepo v1.2.0

Self-contained C# executables, built and verified by the dotnet-binaries workflow. See CHANGELOG.md.

datarepo v1.1.0

Self-contained C# executables, built and verified by the dotnet-binaries workflow. See CHANGELOG.md.

datarepo v1.0.0

Self-contained C# executables, built and verified by the dotnet-binaries workflow. See CHANGELOG.md.

datarepo v0.32.0

Self-contained executables built and verified by the binaries workflow. See CHANGELOG.md.

datarepo 0.31.0

dataRepo turns reanalysed public proteomics data into one repository that AI agents can query and people can browse. This is the first release with downloads that need no Python.

Download

Platform	File
Windows x64	datarepo-0.31.0-windows-x64.zip
Linux x64	datarepo-0.31.0-linux-x64.tar.gz
macOS, Apple silicon	datarepo-0.31.0-macos-arm64.tar.gz
macOS, Intel	datarepo-0.31.0-macos-x64.tar.gz

Unpack the file and run datarepo/datarepo (datarepo\datarepo.exe on Windows). On Linux and macOS, use tar -xzf, which keeps the executable bits. Each download contains Python, DuckDB, pyarrow and pyMzLib with its mzLib bridge, so nothing else needs to be installed.

datarepo/datarepo doctor
datarepo/datarepo ingest  manifest.yaml PXD036557
datarepo/datarepo publish manifest.yaml --site site

macOS: the files are not notarized. If macOS blocks the program, run xattr -dr com.apple.quarantine datarepo once in the unpacked folder.

How these were checked

Each download was built from this tag on its own platform. It then ran the whole getting-started tutorial on a PATH with no Python, and then again with the Python install that built it. The build fails unless both produce the same bundle id and the same catalog id. All four produced bundle 3b56a967d6968d39 and catalog a90e1b47c81f28d5, so Windows, Linux and both macOS builds write identical rows.

datarepo doctor prints the commit an executable was built from.

What's new since 0.28

• datarepo publish: one command that builds the catalog from every ingested dataset and writes the static site.
• Question-neutral output. A site states its research question only when the operator passes --purpose. PXReprise's pxreprise-provenance/1 record is read alongside aging-provenance/2 and /3.
• Curated sample characteristics (aging study layer 0.5.0). Tissue, cell type, cell line and disease read from papers are stored apart from what depositors wrote.
• Documentation rebuilt:
  • a ten-minute tutorial, tested by running it;
  • an operator's guide;
  • a command reference generated from the program.

Details are in CHANGELOG.md.

Upgrading an instance: core schema 0.0.13 is unchanged, so bundles from 0.28 load as they are, with no re-ingest. Deliveries to the aging study layer must now declare layer 0.5.0.

Pre-release software. The schema is not locked yet, and bundles and catalogs are content-addressed, so a change gives new ids rather than altering cited data. Code is MIT-licensed.