robertoostenveld
commented
1 month ago Hi Cyril,
Let me approach this from a general SIESTA perspective, not from medical (neuro)imaging or an ERP perspective.
If you look at the flow in https://github.com/SIESTA-eu/wp15/blob/main/README.md we have input data staging (step 1, 4, 5, 6) and output data staging (step 9 and 10). I would say that the part of the computed results that the data owner considers to be personal cannot be part of the output data that is shared with the researcher. Or phrased differently, the data owner will say "these results here cannot be shared, but the other results can be part of the output data".
For the input data we have a draft process in place that documents which files are available for the computation (namely those files that go through https://github.com/SIESTA-eu/wp15/tree/main/BIDScramble). For the output data we don't have a process yet to document the files that need to be vetted by the data owner. We should make such a process and fit it into the procedure. I think that requires the computation to run on the pseudo data.
I listed all the group level files from usecase-2.4 here
we discussed now sharing subject level but many of those files have subject level information, which I see as necessary to understand the group level results (for most). Rather than make you read the LIMO tools doc, here is an example with MMN:
-1- Having the Yr1 and Yr2 files (any Yr* from LIMO), allows to do further contrasts locally (ok here it's a t-test but say for an ANOVA). We can also not share those, and any contrasts needs to be planned.