Its already on the roadmap, but technical procedure is still undefined. We have this as WP in our next de.NBI funding period.

We can use this issue for discussion. You suggest any repository. This needs to be defined. I doubt that github is the right place because of the size of the data, because data might not be text based and will most likely not changed once it has been deposed.

I have been using Zenodo a lot recently and am very happy so far - diverse datasets and very few issues (>56 datasets and only one failure on one file associated with one that I could solve in the end).
I have not yet tried with larger datasets but it may be worth starting.
Advantage for European sides: the cross-linking with OpenAIRE is nice for our reporting requirements.
The other efficient alternative I could think of is GNPS MassIVE FTP links - we have this for our Swiss Wastewater study. Zenodo gives a DOI and authorship, cross linking and a whole lot more features on the repository side; MassIVE comes with all that GNPS offers if it works on your files (it didn't work on mine tho).

We are also discussing internal possibilities here ... we would have the capabilities but they are not up and running even internally yet for us as a group, our BioCore are running this for others.

I was just thinking about our UFZ repository for our records. However, they still not have DOI and thus the landing page and URL may change at any time....

However, we may firstly indroduce a new tag for the linking or should be use the comment tag?

If this is something we want to try to implement systematically, I would prefer a new tag rather than overloading the COMMENT field with too many diverse comments.
Either e.g. a RAW_DATA entry in the first section, or SP$RAW_DATA?

I would kind of like to use MassIVE or alternatively MetaboLights or MetabolomicsWorkbench so we don't fragment (pun intended) mass spec data over even more repositories. But if Zenodo is getting mass-spectrified already then I guess the argument is not so strong.

Yes to considering MetaboLights and the MetabolomicsWorkbench but I have also heard a lot of feedback that it takes a while to get data into those two ...

Does Metabolights store raw data (meaning ventors' format, not mzML)? Actually, we have to store both vendor and open format.... I will talk to Reza Salek to discuss opportunities.

Yes, it does store raw data formats. Keeva from MetaboLights is also present at the Metabolomics2019 conference.
…
On Tue, 25 Jun 2019, 10:33 Tobias Schulze, @.***> wrote: Does Metabolights store raw data (meaning ventors' format, not mzML)? Actually, we have to store both vendor and open format.... I will talk to Reza Salek to discuss opportunities. — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#79?email_source=notifications&email_token=AARV56FQJWIWBCJFQIWGTZ3P4HJ45A5CNFSM4H23ZJPKYY3PNVWWK3TUL52HS4DFVREXG43VMVBW63LNMVXHJKTDN5WW2ZLOORPWSZGODYPPGWY#issuecomment-505344859>, or mute the thread https://github.com/notifications/unsubscribe-auth/AARV56C4PCLI5K7AEAMTAITP4HJ45ANCNFSM4H23ZJPA .

Thanks for the advice, I already contacted Keeva.

If referencing one or more individual spectra in some repository, the PSI spectrum identifier
could be used, which directly indexes individual scans in raw data:
http://www.psidev.info/usi
we need to find a few examples of prototypes already implementing this.
Yours, Steffen

Sebastian's request was to be able to obtain the input spectrum/spectra used in a record prior to processing.
And MetaboLights can (also) store vendor data, see MTBLS297, Sciex MTBLS2, Bruker and some more if we search for them. Yours, Steffen