v0.2.7

What's Changed

• update: memory efficient to_npz by @jonathanoesterle in #30
• fix: vis2d by @huangziwei in #31

Full Changelog: v0.2.6...v0.2.7

v0.2.6

What's New

Major

Radii Calibration by @jonathanoesterle in #29

This is huge, as one might want to have more accurate neurite radii for a lot of reasons. It might be slow to compute (thus not include in the skeletonization step), but definitely worth doing it in postprocessing. See README and the example notebook for more details.

Minor

• add: plotting options,including rasterize by @jonathanoesterle in #28

Full Changelog: v0.2.5...v0.2.6

v0.2.5

Following up the previous two releases, now we have a better handling of ntype thanks to @jonathanoesterle .

ntype was an afterthought in skeliner. It was not considered in the automatic pipeline originally, and was set to soma=1 / others=3 conveniently, which didn't make too much sense looking back. Later we added post.set_ntype because we need to annotate the axon, but it was not that well integrated into other post-processing functions (e.g. post.prune() ignored the ntype attributes completely, which yields mismatched ntype arrary with the nodes and edges.)

As we added more comprehensive manual post-processing pipeline in v0.2.3, the problem of the previous ntype handling becomes more obvious and prominent. We tried to fix it in this release, and we hope we did :-)

What's Changed

• Change: more sensible default and handling of ntype by @jonathanoesterle in #27

New Contributors

• @jonathanoesterle made their first contribution in #27

Full Changelog: v0.2.4...v0.2.5

v0.2.4

So there was a bug in the manual post-processing pipeline (introduced in v0.2.3), where sometimes there will be two nodes have ntype == 1, because we forgot to check and validate the ntype remapping after soma detection or reroot. The problem was not present in the automatic post-processing pipeline as there's no ntype to remap there.

v0.2.4 fixed this bug. Or, if you somehow stuck at v0.2.3 and don't want to upgrade, you can always just set skel.ntype[:] = 3 after you just got the raw skeleton (without any auto post-processing, of course; don't do it if you used auto post-processing) and before the manual post-processing.

What's Changed

• Fix: ntype after remapping by @huangziwei in #26

Full Changelog: v0.2.3...v0.2.4

v0.2.3

What's Changed

• Refactor: bring post-processing on par with the core process by @huangziwei in #25

This is MAJOR refactoring. The public APIs stay unchanged, only new additions, that is, as the RP title said. This will only break code that use the internals instead of using the public APIs (e.g. pywarper and flatone).

• pywarper<=0.2.3 will only be compatible with skeliner<=0.2.2
• pywarper >=0.2.4 will be compatible with skeliner>=0.2.3 onwards.

if you use the latest versions of all of them, it should be fine. I hope.

See the RP text for more details.

Full Changelog: v0.2.2...v0.2.3

v0.2.2

What's Changed

• distance from points to skeleton surface or centerline by @huangziwei in #24

Full Changelog: v0.2.1...v0.2.2

v0.2.1

What's Changed

• fix: skeletonize() ignored soma_init_guess parameters by @huangziwei in #23

Full Changelog: v0.2.0...v0.2.1

v0.2.0

What's Changed

• v0.2.0 by @huangziwei in #22

This update added compute_contacts_stats() and filter_contact_sites()to skeliner.pair , marked the completion of scope of contact sites detection between a pair of cells (skeletons + meshes).

Full Changelog: v0.1.14...v0.2.0

v0.1.14

What's Changed

• v0.1.14: added post.reroot(); fixed dx._voxelize_union() by @huangziwei in #21

Full Changelog: v0.1.13...v0.1.14

v0.1.13

What's Changed

• v0.1.13: minor refactor by @huangziwei in #20

Full Changelog: v0.1.12...v0.1.13