Final NIH Policy for Data Management and Sharing is out. Settling in with some coffee. https://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-013.html

No strong mandate. I'm mostly okay with this, but boy they better have some administrative governance elsewhere to cover the guaranteed non compliance of awardees.

Definition of data includes data quality sufficient to validate or replicate.

Submission of DMS plans is now required at application, as opposed to just-in-time - this is fantastic news. Just-in-time would have been a giant sunk cost fallacy experiment in action.

Continues to allow significant flexibility across the ICOs. Also good. The NCI isn't NIAID.

"three concepts" to guide sharing human data. 1-use informed consent. 2-communicate limitations to data sharing platforms and stewards (this implies they can limit!). 3-consider what limitations are appropriate, even if not required by law

<#3 here is new and good>

Explicitly allows for more liberal sharing in some forms of data, if consistent with practices/norms/law, and for communities that want to go more aggressively open. Also good.

Addresses AI/AN concerns here. I'd like to see more teeth, and explicit, dedicated funding for research emerging specifically from AI/AN communities. DMS plans are way more likely to be good for those communities if they come from those communities.

“[s]hared scientific data should be made accessible as soon as possible, and no later than the time of an associated publication, or the end of the award/support period, whichever comes first.”

i love dependent clauses that orient in the correct direction

"How Long Data Should Be Available"

"Where to Share Scientific Data" - established repositories, no change. Don't roll your own!

Also hey check out http://synapse.org  we subsidize most storage and transfer fees, and have extensive access control / governance you can inherit

Sticking with the 2 page limit. And although I'm all about that semantic life, thank god they didn't require it in the DMS plans themselves. Better to have a scientist thinking about the actual DMS plan than puzzling through a contributor role ontology.

"The final Supplemental Information eliminates the language that a response of “to be determined” is acceptable."

"The final Supplemental Information asks researchers to describe how the scientific data will be findable and identifiable, i.e., via a persistent unique identifier or other standard indexing tools. "

Explicit callout to the F and the I parts of FAIR, nice.

Moving security responsibility to the platforms. OK. Can you give some clarity to us platforms, please? We're over here assuming we have to be around HIPAA or HITRUST or FISMA medium. Please advise.

"Personnel costs required to perform the types of data management and sharing activities described in the final Supplemental Information are allowable."

Clarity is good! Also, you can pre-pay long term storage fees, but have to use the $$ during the grant cycle.

"NIH does not intend to provide a comprehensive list of suitable repositories outside of those supported or stewarded by NIH."

They will instead release a supplemental doc with advice on how to evaluate a repository. I don't see it linked in the policy itself although it is described in the present tense.

Effective date: January 25, 2023

(I know we have to give people time, but two years seems a bit excessive)

And that's all I got for now. I'll add to this thread when I see the supplemental on how to select a repository.

Via @DNAlawyer - the supplemental is at https://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-016.html

- Repeats that ICOs can be quite specific in FOA process. - - Encourages choosing a discipline specific repo
- Datasets under 2GB can be attached to manuscripts at PMC.
- Calls out cloud preservation options (again, that's http://synapse.org )

Desiderata: PIDs, sustainability, metadata, curation/qa (hey! awesome!), free and easy access, broad and measured reuse (results will be a downer, but good to measure), use guidance, security/confidentiality, common format, provenance, retention policy.

FIDELITY TO CONSENT

(I know that seems obvious, but whooo boy there is let's say a gap between theory and practice in this one)

Also kind of big here is the "even if de-identified" - no more scooting through that loophole.

OK now i'm really done. Overall, kudos to NIH office of science policy on this. Not an easy thing to shepherd through to final. My complaints are far less than the sum of good stuff here.

Time now to teach the grants offices how to implement and hold people's feet to the fire.