Download everything as one file

The whole public dataset as a single zip: every published figure, plus manifest.csv naming each one.

This endpoint answers 302. The archive lives in object storage, so the redirect points at a CDN URL that supports range requests and resuming - no bytes pass through the application. Any ordinary HTTP client follows it; curl needs -L.

The archive is rebuilt when the dataset changes, and its URL carries the dataset version, so a URL you have already fetched is safe to cache indefinitely. Ask this endpoint again to learn the current one.

503 means a figure has been published since the last build and the archive for this version does not exist yet. Nothing is missing - the manifest is complete throughout, and the reply names it.

Add ?gene= to get a zip of named genes instead, built on request. A key scoped to part of the dataset always gets a zip built this way rather than a redirect, because the shared archive holds every supplier.