Is there a timeline for when dandisets with Zarr assets can be published?

[joshlawrimore]

I want to host derivative images of light sheet microscopy scans of intact mouse brain packaged as ome-zarr using the BIDS-microscopy standard. I noticed the Publishing Dandisets section of the handbook states "NOTE: Dandisets with Zarr assets currently cannot be published. We are actively working on enabling this feature."

1. Does this mean data packaged in ome-zarr cannot be assigned a DOI at all?
2. Is there a timeline as to when datasets containing Zarr assets can be published?

[yarikoptic]

1. yes
2. no specific date. We are ATM still evaluating https://earthmover.io/blog/icechunk and deciding on alternatively to proceed with custom implementation.

And alternative path, if the goal is just to get a DOI (not necessarily pointing to a specific version) could be for us to prioritize dandi/dandi-archive#2012 design/implementation and thus provide DOIs even for draft dandisets.

What is your ultimate goal behind getting a DOI -- immutability or just citeability?

[yarikoptic]

I think we finished draft doi design doc. Might approach implementation next week or early next year.

[joshlawrimore]

Has there been any progress on getting DOIs for Dandisets containing ome-zarr assets?

[yarikoptic]

progress - yes. result - not yet. Realistically, I think - within a month we should get "real DOIs" going for dandisets. But one discovery is that unless they are non-draft, they would kinda be useless. Adding a note: https://github.com/dandi/dandi-archive/pull/2012/files#r1926301678 .

[agt24]

Thanks @yarikoptic

Not 100% sure I follow the concerns about useless DOIs but I think DOIs on non-draft DandiSets are OK for our use case as long as they are immutable (+/- versioning ala Zenodo) and will be eventually citable.

[yarikoptic]

That's the catch, zarrs are mutable ATM. Dois will be citable though.