Read MIT Annotations

[withmywoessner]

Reference issue (if any)

Fixes #12660.

What does this implement/fix?

Adds the ability to read WaveForm Database (wfdb) annotations using the already existing library. My proposed API adds two new parameters to mne.read_annotations() fmt and suffix. fmt and suffix are used in combination to read an annotation whose file suffix is not strictly defined.
If mne.annotations(fmt='wfbd'...) is called with no suffix, then the function will check to see if the suffix matches a default list of suffixes.
If mne.annotations(fmt='wfbd', suffix='seizures'...) is called, then the function will bypass the default list of suffixes.

[withmywoessner]

Hi @drammock and @larsoner, I know you mentioned reading the annotations using a predefined list of suffixes in #12660, but I thought it would be useful to also allow for the option to read files using any suffix. Let me know if you prefer otherwise.

[withmywoessner]

Where is the best place to list the suffix mappings @larsoner? The WFBD library mentions a few that I would also like to include besides .seisuzes.

[larsoner]

Where is the best place to list the suffix mappings @larsoner?

Looks like the mapping currently lives here

mne-python/mne/annotations.py

Line 1203 in b38385e

readers = {

[withmywoessner]

Okay, everything should be good then. Should I add tests and include a citation for the library somewhere? @larsoner

[larsoner]

Should I add tests and include a citation for the library somewhere?

Sure, it would be good to add it to environment.yml (and to conda-forge if it's not already on there) and pyproject.toml in the full deps. Then we'll also want to add it to mne-installers around https://github.com/mne-tools/mne-installers/blob/a18bf50444a0e258b7d09c0ce14eab1ec8d948a9/recipes/mne-python/construct.yaml#L92

[withmywoessner]

Would you mind cutting a new release for the testing data? @larsoner
mne-tools/mne-testing-data#118

I used the data from https://physionet.org/content/chbmit/1.0.0/

Then we'll also want to add it to mne-installers around https://github.com/mne-tools/mne-installers/blob/a18bf50444a0e258b7d09c0ce14eab1ec8d948a9/recipes/mne-python/construct.yaml#L92

Added a PR here:
mne-tools/mne-installers#325

[larsoner]

https://github.com/mne-tools/mne-testing-data/releases/tag/0.157

[withmywoessner]

Can anyone help me figure out why the new test file is not being recognized by pytest? Is it because it has two extensions? When I read it myself it works fine.

[drammock]

Can anyone help me figure out why the new test file is not being recognized by pytest? Is it because it has two extensions? When I read it myself it works fine.

You need to tell the CIs (via MNE-Python) to actually use the newest version of the testing dataset. See:

https://github.com/mne-tools/mne-python/pull/12800/files

[withmywoessner]

I see, thank you. The testing data PR was merged before I updated the version.txt file. Should I make a new PR or ask for the last one to be reverted?

[larsoner]

Whoops I missed that! Will fix both and cut a new version

[larsoner]

Done... use 0.158 🤦

[larsoner]

@withmywoessner do you have time to come back to this or would it help if someone pushed commits?

[withmywoessner]

Yes, I can do this! Sorry about that! I have been dealing with NIH nonsense this year.