Don't perform ICA cleaning on raw data by default when epochs are present #926
Conversation
|
I'm not sure this is the right approach (in terms of, where to do the conditional branching)… @larsoner Could you kindly take a look? |
| - When using ICA or SSP with resting-state data, we now automatically produce cleaned raw data files. This | ||
| behavior can be controlled via the new [`spatial_filter_raw`][mne_bids_pipeline._config.spatial_filter_raw] | ||
| configuration option. (#840, #926 by @larsoner and @hoechenberger) |
There was a problem hiding this comment.
The introduction of writing raw data was actually in 1.6 / #840. So the behavior change here is actually to stop writing files by default. So maybe we should leave the default as True for backward compat. Then this PR is just to add the ability to have False | None.
Personally I think True is the most reasonable default anyway. If you ever want to re-epoch your data (I often do!) then you'll want those preprocessed raw files. And you can cross-check individual runs against the produced epochs, it's more complete to have raw data report at each stage, etc. even though it makes the report longer.
|
Thanks for your help here, @larsoner, this is much appreciated! I'd like to hear @SophieHerbst's input as well, esp. since she said they were struggling with disk space usage and all… |
Fixes #925
Before merging …
docs/source/changes.md)