Repository navigation
Offer: move data to Zenodo with BiocFileCache caching, as just done for curatedCRCData and curatedOvarianData #2
Description
Hi Markus,
I just moved the data out of curatedCRCData and curatedOvarianData to Zenodo, prompted by both packages being past the new limit imposed by Bioconductor's shift to R-universe (100 MB) (waldronlab/curatedCRCData#4, waldronlab/curatedOvarianData#9). Happy to send you a PR doing the same for curatedBladderData — I could even do the Zenodo upload under my account, or you could, whichever you prefer.
What it looks like (see waldronlab/curatedCRCData#5 and waldronlab/curatedOvarianData#10 for the merged PRs):
- Each ExpressionSet goes on Zenodo as its own xz-compressed .rda; a new getter (
curatedBladderData("GSE89_eset")) downloads on first use and caches with BiocFileCache, md5-verified against a manifest ininst/extdata/. data(X)keeps working through smalldata/*.Rstubs that delegate to the getter with a deprecation message, so nothing breaks for existing users. During the package's own R CMD check the stubs resolve to small bundled subsets, so the builders never touch the network.- The vignette and
createEsetList.Rrun offline on those subsets; tarball drops from hundreds of MB to a few MB.
Both packages pass R CMD check and BiocCheck cleanly with this, and an integrity report verifies every Zenodo object is all.equal to what the package shipped before.
Any interest? If so I'll put the PR together — the changes are mechanical at this point.
-Levi (PS - easy to do thanks to Claude, including the above text :D)
🤖 Drafted with Claude Code
Activity
lima1 commentedon Oct 6, 2026
Hey Levi, sounds great, for sure I’ll accept a PR. I don’t remember using Zenodo, so if you could host, that would be great. PS: I’m a semi-retired dinosaur without Claude (yet?).
lwaldron commentedon Oct 6, 2026
PR is up: #3. Data is on Zenodo at https://zenodo.org/records/23192472, deposited under my account as you suggested. Happy to change the creator list or license on the record if you'd prefer something different.
Tarball goes from about 75 MB to 1.4 MB.
data(GSE89_eset)still works through small stub files, so existing scripts and the reverse dependencies don't break.One thing worth knowing: your GitHub copy was at 1.19.2 while Bioconductor devel is at 1.49.0, since the release process bumps versions on the Bioconductor git without touching GitHub. The two are otherwise byte-identical, so the PR is based on your GitHub tree with the version set to 1.49.1.
I also fixed a few unrelated things I ran into (biocViews listed OvarianCancerData, the package help page described ovarian cancer with the wrong author list, affy was unused). They're listed separately at the bottom of the PR and easy to drop if you'd rather keep this focused.
🤖 Drafted with Claude Code