Export to EEGLAB format?

[laemtl]

I try to write a raw object using

raw = mne.io.read_raw_eeglab(input_fname=file, preload=False, verbose=True)
write_raw_bids(raw, bids_basename, allow_preload=False, overwrite=False, verbose=False)

and I got the following error:

Cannot write file - The data has already been loaded from disk. To write it to BIDS, pass "allow_preload=True" and the "format" parameter.

When I set allow_preload to True, I then get
For preloaded data, you must set the "format" parameter to one of: BrainVision, EDF, or FIF

The issue is I don't see an option of EEGLAB/set file.

Is this a know issue? Am I doing something wrong?

[hoechenberger]

I don't think we support writing to EEGLAB. You can try to write to BrainVision instead

[agramfort]

what is the content of bids_basename and bids_basename.format ?

…

Message ID: ***@***.***>

[hoechenberger]

For preloaded data, you must set the "format" parameter to one of: BrainVision, EDF, or FIF

You will have to pick one of those. No EEGLAB support so far. Sorry!

[sappelhoff]

We can consider adding support for exporting to EEGLAB, using this library: https://github.com/jackz314/eeglabio

[christinerogers]

Thanks @sappelhoff
Full support for .SET would help us make open-source BIDS converter tools like EEG2BIDS accessible for many more BIDS-compliant cases, especially for researchers looking to move away from manufacturer-driven data formats like BV.

(Let us know if there's a separate issue to follow up on for this, my colleague @laemtl is raising this somewhat related also to #1151 )

[sappelhoff]

Full support for .SET would help us make open-source BIDS converter tools like EEG2BIDS accessible for many more BIDS-compliant cases, especially for researchers looking to move away from manufacturer-driven data formats like BV.

3 comments on this:

1. The EEGLAB data format is not "RECOMMENDED" under BIDS, it's merely "accepted" based on its wide distribution
2. it's a MATLAB file in disguise
3. The mne-bids devs (to my knowledge) don't work actively with the EEGLAB format themselves

that maybe explains why so far there have been only little efforts to support this file format.

Having that said, we'd welcome contributions in the form of PRs (perhaps by @laemtl) to allow export to EEGLAB.

What do you think @hoechenberger @agramfort ?

[hoechenberger]

Isn't Matlab simply HDF5 these days? Hence I don't see why (modern) EEGLAB shouldn't be fully supported by BIDS (and hence MNE-BIDS)

[sappelhoff]

@christinerogers @laemtl @jeffersoncasimir

Please see this issue as a list of To Dos as to improve EEGLAB functionality of mne-bids:

• Enhance copyfile_eeglab #654

We'd be happy to review PRs to improve the state of the art. And perhaps working on these "smaller" items can be done in parallel to the present issue of also exporting to the EEGLAB format.

FYI, I have created a project board for all EEGLAB related issues: https://github.com/mne-tools/mne-bids/projects/10#card-89935503