MNE python features wish list

• noise covariance automatic regularization using sklearn.covariance.ledoit_wolf
• add deep structure to source space using freesurfer aseg volume segmentation
• multivariate functional connectivity measures such as partial directed coherence
• support IO with more file formats (eg. EEG brainvision files and CTF data)
• project mixed sensor MEG data to virtual magnetometers like mne_analyze. Useful for topograhy with neuromag data.
• native forward solution calculation