Review of example BEP032 datasets

[robertoostenveld]

@JuliaSprenger has prepared a number of examples that are located here https://gin.g-node.org/NeuralEnsemble. These are now ready for review, please post your comments here.

[bendichter]

In the Petersen NWB example, there is a mention of cooling of the brain dynamically over time, but I cannot find a signal for this.

I know from the corresponding dandiset that there is also a signal of the position of the animal over time, which is also missing, and necessary for reanalysis.

The data files are very small. Presumably they are in "stub" mode, which is just the first few timestamps of the voltage series. That's fine to demonstrate the BIDS structure, but might not be sufficient if we are thinking about packaging a dataset for reuse.

[jcolomb]

looking at the neuropixel example:

• https://gin.g-node.org/NeuralEnsemble/BEP032-examples/src/master/ephys_neuropixel/code/create_rawdata.py : not in git, not in git annex ? Note it is still in the history at https://gin.g-node.org/NeuralEnsemble/BEP032-examples/src/225bb5d6a511530a9945765a0f82d8ea6c254ad7/ephys_neuropixel/code/create_rawdata.py

• I am not an expert (yet), but it looks to me that it would be interesting to see how the derivatives data would look like in that example. Due to the enormous size of the raw data, we will probably see the raw data archived and only the derivative data (after spike sorting) published.

• By the way, is there a way to make that (raw data not available) BIDS compatible ?

• probably outside of this single example purpose, but the https://gin.g-node.org/NeuralEnsemble/BEP032-examples/src/master/ephys_neuropixel/rawdata/participants.tsv file does not give much information.

• wondering where information about probe position would end up ?

[SylvainTakerkart]

In the Petersen NWB example, there is a mention of cooling of the brain dynamically over time, but I cannot find a signal for this.

I know from the corresponding dandiset that there is also a signal of the position of the animal over time, which is also missing, and necessary for reanalysis.

The data files are very small. Presumably they are in "stub" mode, which is just the first few timestamps of the voltage series. That's fine to demonstrate the BIDS structure, but might not be sufficient if we are thinking about packaging a dataset for reuse.

Hi @bendichter , thanks for your feedback! This data set was initiated a long time ago, never really finalized and with @JuliaSprenger , we were planning not to include it (mostly because of lack of time on our side, and because we were planning to have the neuropixel data set as a nwb example data set). However, if somebody in the nwb crew has time to finalize its BIDSification, that'd be great! And / or if somebody in the dandy/nwb crew has time to write some kind of dandy2bids converter, that would also be a great addition (even if it does not work in 100% of the cases...). From what @JuliaSprenger told me recently, maybe @oruebel could have ideas about this?