[Mne_analysis] Making SSP files with python

[Alexandre Gramfort]

hi jon,

> I've been working with the python tools to build SSP projectors for EOG and
> ECG.   I had thought that it might be necessary to use make_projector after
> compute_spatial_vectors, but write_proj seems to work perfectly well with
> the output from compute_spatial_vectors.

the make_projector function assembles the projection matrix while the structure
returned by compute_spatial_vectors contains what is stored in a fif files.
So what you describe is correct.

> The output files are usable in both
> MNE and Neuromag.  As with projectors build using mne_process_raw, xfit
> treats these projectors as if they had been applied explicitly to the data.
>  Using the sample data, what I'm doing is:
>
> raw_fname = data_path + '/MEG/sample/sample_audvis_raw.fif'
> raw = fiff.Raw(raw_fname)
> exclude = raw.info['bads'] + ['MEG 2443', 'EEG 053']
>
> # Make list of channel names into an array suitable for index
> names = np.array(raw.info['ch_names'], dtype='S7')
>
> # Do EOG
> event_id = 998
> eog_events = mne.artifacts.find_eog_events(raw, event_id)
> picks = fiff.pick_types(raw.info, meg=True, eeg=False, eog=False, ecg=False,
> stim=False, exclude=exclude)
> ch_names = names[picks]
> tmin, tmax = -0.5, 0.5
> epochs = mne.Epochs(raw, eog_events, event_id, tmin, tmax,
> baseline=(-0.5,-0.2), picks=picks, proj=False)
> data = epochs.get_data()
> EOGprojs = fiff.proj.compute_spatial_vectors(epochs, n_grad=1, n_mag=1,
> n_eeg=0)
>
> EOGprojfile =  raw_fname[:-4] + '_EOG_auto-proj.fif'
> fid = fiff.write.start_file(EOGprojfile)
> fiff.proj.write_proj(fid, EOGprojs)
> fiff.write.end_file(fid)
>
> Am I skipping a step somewhere along the way, or does this seem reasonable?

yes it does.

we might however need to do something so you don't need to manually
start end end the fif files.

if you think it could be written in a smaller/cleaner way let me know.

Alex