preprocessing copies

[mangstad]

I've been meaning to fix this for awhile but we need to edit the preprocessing central script to make a copy of anatomicals so coregistration doesn't modify the headers. even though this is expected behavior from the current version this should probably be classified as a bug.

Sent via Hubroid

[dankessler]

While we're at it, could we maybe run it through this logic

1. If a NIFTI file, convert to img/hdr pair
2. Copy hdr, soft link to img

That way, we don't duplicate all of our anatomical data unnecessarily every time we preprocess? What do you think @mangstad ?

[mangstad]

It's certainly possible, but I'm not sold on it. I think space savings would be minimal, as there are much bigger contributing factors to our excessive space consumption than having a few extra anatomy files around (namely having multiply duplicated functional runs). The added complexity of having a bunch of half sym-linked file pairs that can end up pointed nowhere if paths change or folders are reorganized is probably not justified by the space savings.

[dankessler]

Fair enough, maybe it's not worth the risk to save a bit of disk space

[mangstad]

Looks like this can be taken off the milestone since we're not actually releasing the matlab-based preprocessing, but it is still an issue that needs to be addressed for internal use.

[dankessler]

Cool. This is nice having other people be able to adjust tags, milestones, etc :)

[dankessler]

@mangstad do you think we could take care of this prior to next release?

[mangstad]

Honestly, since we're not officially releasing the matlab-based preprocessing we could probably just close this -- or at least unclassify it as a bug.

[rcwelsh]

is this an issue anymore since we are trying to move people to spm8Batch -- which does copy files to prevent them from being modified?