I am curious to know how other people/institutions may be planning for NIH's
new data sharing policy for all proposals submitted to them after 1/25/2023,
particularly as it affects NMR data. In a nutshell, the policy states "shared
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.” PI's have to include an acceptable plan
for how to do that. The data is to be made available whether or not it gets
included in a publication, or whether it supports a finding. Certain types of
research materials do not have to be released, for instance research notebooks,
but as far as I can tell NMR data would not be one of the exempt categories.
See: https://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-013.html
NOT-OD-21-013: Final NIH Policy for Data Management and Sharing<https://grants.nih.gov/grants/guide/notice-files/NOT-OD-21-013.html>
NOT-OD-22-029 - Request for Information on Proposed Updates and Long-Term
Considerations for the NIH Genomic Data Sharing Policy. NOT-HG-21-023 -
Notice Announcing NHGRI Guidance for Third-Party Involvement in Extramural
Research. NOT-HG-21-022 - Notice Announcing the National Human Genome Research
Institute’s Expectation for Sharing Quality Metadata and Phenotypic Data
grants.nih.gov
In my facility, NIH grants pay for thousands of spectra per year, so it is
by no means too early to think about how to support whatever data sharing plans
PI's might submit to NIH, and how to automate whatever steps are involved in
the actual data management. And it might not hurt to test a provisional plan
in advance, rather than get a few years into it, and decide plan A isn't
really workable and it all has to be implemented again in plan B ...
Are we thinking that a PDF of an NMR spectrum is an acceptable way of making
the data accessible? Or at long last, should we provide access to FID's?
If the latter, how to tie the spectrum to metadata, like what it's a spectrum
of? (Right now, if I were to try to make head or tails out of an older spectrum
with an inscrutable name like XYZ-VII-209, I'd likely request to view the
person's notebook, but that won't be part of the accessible record.) Should
we recommend MNova or ACD documents, which might contain mixed types of
analytical data and metadata? NMReDATA or other markup files?
The new policy also covers that the data is supposed to be of sufficient
quality to be fit for purpose. How should that be defined and evaluated,
ideally in an automated way, and close to real time?
Personally, I think change is wildly overdue on this, but the more ambitious
the plans are for sharing data, the more complex it will be to do it successfully.
David VanderVelde
Manager, Liquids NMR Facility, Caltech
davidv_at_caltech.edu
Received on Mon Jan 10 2022 - 15:28:34 MST