problem with sample data set: attempt to compute noise-cov fails

Hi,

I am trying to review the tutorial in Chapter 12. In section 12.10 I entered the mne-process-raw command to compute the noise-cov matrix,

mne_process_raw --raw sample_audvis_raw.fif --lowpass 40 --projon --savecovtag -cov --cov audvis.cov

and got this error:

Too few samples (required : 1835 got : 363)

I used fresh copies of the input files.

What could be going wrong? I cannot continue with mne-do-inverse-operator since it complains it does not have the sample-aaudvis-cov.fif file.

Thanks,
-Jeff

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20120413/10b52f28/attachment.html
-------------- next part --------------
A non-text attachment was scrubbed...
Name: noise-cov error.rtf
Type: text/rtf
Size: 4293 bytes
Desc: noise-cov error.rtf
Url : http://mail.nmr.mgh.harvard.edu/pipermail/mne_analysis/attachments/20120413/10b52f28/attachment.rtf

It's likely that you either have too few events being used, or there
are too many epochs that are being thrown out (due to exceeded
thresholds or flat channels).

Eric

Eric,

Thanks for the input, but this is part of the sample/tutorial data
provided with the MNE installation and manual, and should just work. I am
hoping someone at MGH will respond.

-Jeff

Hi Jeff,

Eric is right the Tutorial tries to teach you what you need to do with real data, so there are bad channels that need to be marked in the tutorial data set. The manual states:

Mark the channels you identified in step 6. bad for this viewing session by clicking on their channel names on the left. You can save the bad channel selection to the file from File/Apply bad channels . Bad channel marking can be removed by clicking on their channel names again and selecting File/Apply bad channels . Alternatively, you can use the utility mne_mark_bad_channels to set a bad channel selection, see Designating bad channels: mne_mark_bad_channels.

This must be done in order to successfully calculate averages and covariances.

D

Hi Jeff,

as Dan said the raw sample data keeps a few steps for the new user to
do manually in order to learn:
mark bad channels, define rejection thresholds, filtering parameters
and coregistration.

if you want to play with the preprocessed version of these data see:

ftp://surfer.nmr.mgh.harvard.edu/pub/data/MNE-sample-data-processed.tar.gz

FYI that's the data used to illustrate all the Python examples.

Best,
Alex