Hello,

Along the same ideas than what you are talking about, nilearn provides a few data downloaders:

http://nilearn.github.io/building_blocks...

Our goal is not to be exhaustive, but rather to fetch a few dataset well. Also, we prefer fetching datasets that have been preprocessed.

We do have some very nice value in there: the downloaders can resume after a partial download, and they have a simple but handy progress bar.

However, if you want to add a downloader to a preprocessed data, we might be interested. Just open an issue an the tracker to discuss this with the team.

Cheers,

Gaël