can we teach xform_brain to mirror as well

[jefferis]

I was thinking maybe a mirror param that takes a brain. Then using the new nat::reglist format we could combine that registration with the bridging sequence.

@ajdm what do you think?

[jefferis]

would have to have a mirroring reglist that did both the flip along one axis and the non-rigid part. This happens inside nat::mirror but would not be so easy to extract that logic. An alternative might be to have an on disk object called something like XXXX_fullmirror.list or XXXX_fullmirror.rds where the full distinguishes this from a mirroring registration that only corrects the nonrigid displacement.

[jefferis]

Another option might be to ensure that there are always pairs of CMTK registrations that enable flipping and non-rigid correction after mirroring (e.g. JFRC2_imgflip.list and JFRC2_mirror.list but rarely present at the moment).

[jefferis]

Another option. Use a call like this:

xform_brain(sample=A, reference=A, via=mirror_brain(B))

would need to extract the identity of B from the call for this to work though.

[schlegelp]

I was about to open a separate issue for this but related to what you suggest here: could we have mirror_brain to accept a via parameter? For example to mirror FAFB neurons:

mirror_brain(fafb.neuron, brain=FAFB14, via=FCWB)

[jefferis]

I think this is what you want for now. http://natverse.org/nat.jrcbrains/reference/mirror_fafb.html. Someday I'll generalise.

[jefferis]

m=nat.jrcbrains::mirror_fafb(catmaid::catmaid_parse_url('https://neuropil.janelia.org/tracing/fafb/v14/?pid=1&zp=207080&yp=154009.72&xp=390555.66&tool=tracingtool&active_skeleton_id=556874&active_node_id=7135535&sid0=5&s0=1'))
elmr::open_fafb(m)

> elmr::open_fafb(m, open = F)
[1] "https://neuropil.janelia.org/tracing/fafb/v14/?pid=1&zp=186027&yp=162727&xp=671463&tool=tracingtool&sid0=5&s0=1.000000"