[MAINT] Update to latest MNE version #66
Conversation
|
Uh oh, looks like no tests. From here, I will make a minimally reproducible example from the data that Andy wanted to download and possibly implement it as a test. |
|
Ok I looked around a bit, and it looks like the GitHub Actions are out-of-date. Andy expressed interest in updating this during a Sprint, which I think fixing some of the code aspects wouldn't be too hard but the GitHub actions stuff really involves a lot of setup. Andy, you could try copying things over from a currently-maintained package like |
|
Hmmm okay, I gave it a quick try migrating the tests over to a more modern framework so that it could potentially be updated but I ran into a bit of a roadblock because I couldn't get the testing data from |
|
@alexrockhill Yes, I would actually love to do that. Actually, I don't need to wait until the sprint, I can start looking at it now. It is an interesting project for me to learn how it works in general. I might spend an hour or so a day to start looking at it. I am quite new to contributing to an open source project, although I have been using Github for a long time for collaborations within my lab. So I might be a bit slow at the beginning trying to figure things out, e.g., how to setup tests and Github Actions. I will let you know if I have any questions. I will probably attend another MNE office hours to discuss more about this. In the meantime, if you have any good resources that I can go to and read more about this, it would be great. Thank you! |
|
@agramfort, do you have a copy of the testing data for this repo? It doesn't seem like too much work to get it up-to-date and it would be a shame not to try and maintain Dennis's work at least a bit. I haven't set up |
|
... I'd need to dig somewhere. I don't remember well if there was even a script somewhere to generate the testing data. Unfortunately, they are not on my current computer, I will need to do some archival work. |
If you had a minute to do that, it would be super helpful! You could maybe just get it off of AWS with |
|
won't be a minute unfortunately :( but I'll look into it within next week. |
|
@alexrockhill Here's the post that I intended to post on the MNE forum to reply your response (https://mne.discourse.group/t/hcp-data-compatibility/4433/5). As you suggested, it might be good to move our conversation here: The HCP data for one subject is really large. The one for testing that I have on my laptop is more than 50G, and I am not sure how I can share this with you. Basically, I just follow the tutorial here to extract the anatomy from the HCP data: https://mne.tools/mne-hcp/auto_examples/make_mne_anatomy.html#sphx-glr-auto-examples-make-mne-anatomy-py The MNE-HCP seems to provide a good API for running source localization: https://mne.tools/mne-hcp/auto_tutorials/plot_compute_evoked_inverse_solution.html#sphx-glr-auto-tutorials-plot-compute-evoked-inverse-solution-py And they also provide their own function of plotting the co-registration: https://mne.tools/mne-hcp/auto_examples/plot_coregistration.html#sphx-glr-auto-examples-plot-coregistration-py However, they mentioned that for various reasons, they didn't use the MNE coordinate for the source localization, but they didn't mention explicitly why. And this creates some problem while visualizing the HCP data on MNE: https://mne.tools/mne-hcp/generated/hcp.preprocessing.map_ch_coords_to_mne.html#hcp.preprocessing.map_ch_coords_to_mne I am trying to study the source code of MNE-HCP and understand how the process is done. I might attach some code later if I found them useful to clarify. I am curious about why they didn't choose to use the MNE coordinate in the first place? It might be a good clarification question for Denis, but do let me know if you know anything about this. |
|
Yeah, I haven't used HCP data so I have no idea about choosing coordinate frames but I'm happy to work through examples with you. The example data Dennis shared is ~400 MB so that'll be what we want to work with. I assume he downloaded an entire dataset and cropped everything down as short as possible so that you get all the formatting without huge file storage. |
Sounds good. I will come to the MNE office hour this Friday. I think we can take a look at it together then. |
|
So I think the resolution was that nothing actually needs to be fixed, the incorrect orientation was by design because HCP puts everything in BTI coordinates and those are ALS not RAS. Closing this, as I don't think it's necessary to update the tests seeing as we can't have local access to the data, they should just stay as is. If you want to walk though how to do everything in BTI coordinates next office hours, I'd be happy to help with that @andylikescodes |
|
FYI, I started an update of the package here to get it compatible with recent MNE versions: #71 |
Related to https://mne.discourse.group/t/hcp-data-compatibility/4433.
MNE-HCP isn't maintained rigorously but Andy (sorry don't know your GitHub handle) is interested.
I thought I would get the ball rolling with a PR for an example.
First, I'm just adding an empty commit (usually, you would make you changes and then do
git add -Afor all the files or add each changed file individually fromgit statusand then dogit commit -m "your message goes here about what you changed") to see if the existing mne-hcp tests will point out the failures that we'll need to change.