Skip to content
OpenReadout

export-dataset

export-dataset exports the data sets that a search query selects from an index as an ML-ready dataset: images as OME-Zarr, tables, traces and spectra as Parquet (or CSV), metadata JSON and a datasheet. It is resumable: a rerun continues where the last one stopped. Every file is verified.

openreadout export-dataset [OPTIONS] <INDEX_DIR> <QUERY> <OUT>

The query has the syntax of search; "" selects everything.

Flags

  • --limit N: most data sets exported. Default: all.
  • --tables parquet|csv: file format for tables, traces and spectra. Default parquet.
  • --redact: replace values flagged as personal data with stable salted hashes.
  • --salt-file FILE: file holding the redaction salt (or set OPENREADOUT_REDACT_SALT).
  • --license SPDX: license of the whole export. Default: the nearest LICENSE file of each source.
  • --no-images: leave images out.
  • --json: print the JSON wrapper instead of text.

Example

Terminal window
openreadout export-dataset /lab/index "sample~A12" dataset/ --redact --salt-file ~/.salt

Redaction and the personal-data rules: Personal data.

JSON

export-dataset.

Run openreadout export-dataset --help for the full help of your installed version.