Mainak Jas

But wouldn't it be easy to just edit the .bidsignore? I can see why it can be useful to hardcode it in the code, but this could easily become a...

> 1. Should the validator ignore .bak files Thinking again, I would say yes. The validator should not treat files inside .ds separately because it's a multifile system whose [internals...

> If you are talking about any folder and any file I disagree. An empty file such as `sub-01_T1w.nii.gz` is definitely a sign of something going wrong with the dataset...

the ransac implementation in autoreject is mostly there because it's what was used in the paper for comparison. Thus for reproducibility purposes. I haven't tried to optimize it in any...

it's been 3-4 years since I looked at the code, so not sure I can suggest a quick hack. Maybe use this function: https://mne.tools/stable/generated/mne.make_fixed_length_epochs.html#mne.make_fixed_length_epochs ?

> autoreject does seem closer to matlab at least for sub 007 I am pretty sure I checked that the matrices are **exactly** the same as in Matlab, at least...

see the seed here: https://github.com/VisLab/EEG-Clean-Tools/blob/3ed337e83dfaaad2c8e3ecb671a44a21c5b288c0/PrepPipeline/utilities/findNoisyChannels.m#L333 I think I removed it in the final version after checking that it agrees if you place it in exactly the same location (so that...

another source of difference can be the interpolation function itself. Looking at the commits now, I remember that I made this pull request in MNE: https://github.com/mne-tools/mne-python/pull/2758/files to make it exactly...

here's the diff of the pull request I made in a private repository: https://gist.github.com/jasmainak/d459dd128ded1e93b54050c7fab08e79. It has the script of reading the EEGLAB dataset I used to compare. Not sure if...

Does Pyprep also work for MEG data? The reason for Autoreject copying over some of the MNE interpolation code is for efficiency. In the autoreject algorithm, you have to do...