CAVEconnectome / MeshParty

Apache License 2.0
35 stars 16 forks source link

Request: Split `MeshParty` into several packages #91

Open caniko opened 1 year ago

caniko commented 1 year ago

Can we split the skeletonization measures into another package? This package has a lot of bloat around what is otherwise a very useful tool. I am developing meshemy to manipulate meshes in Python, it is still a work in progress, but it provides mesh repair among other functionalities that you provide in MeshParty in addition to the skeletonization.

I will use the source of the skeletonization for now, and hope you split it into something of its own in the future. Thanks.

To be playful with words: A package shouldn't be a party, it should have a well-defined scope.

ceesem commented 1 year ago

We’ve been discussing this for a while, and agree in general. The goal of this tool has moved away from neuronal meshes and small mesh cutouts towards morphological analysis of neurons broadly, generally within the CAVE ecosystem. There is some baggage that’s come along in large part due to the evolution of that ecosystem and the standard types of analysis. However, that scope comprises a lot of features that may seem like bloat to you naturally coexist to us.

Is the part of meshparty you’re finding you would like to be it’s own package the graph-based TEASAR implementation? As we think about a major refactor along these lines, I’d be curious what other parts, if any, you found useful for your purposes.

caniko commented 1 year ago

Thank you for the informal response!

However, that scope comprises a lot of features that may seem like bloat to you naturally coexist to us.

Which I totally understand, I was speaking from my own perspective. I am not interested in the CAVE ecosystem as I find it limiting for the scope of my project, and future ideas.

Is the part of meshparty you’re finding you would like to be it’s own package the graph-based TEASAR implementation?

Yes, along with the add-ons for the modeling of neuron and neurites.

I’d be curious what other parts, if any, you found useful for your purposes.

I did find your animation pipeline intriguing; however, I prefer to explore possibilities through bpy -> Blender. Just the TEASAR implementation.

fcollman commented 1 year ago

I started this process! https://github.com/AllenInstitute/PyGraphTEASAR I have not finished fully testing it on real neurons, but if you wanted to contribute there that would be useful!

fcollman commented 1 year ago

This repo has only numpy and scipy as dependancies.

caniko commented 1 year ago

I started this process! https://github.com/AllenInstitute/PyGraphTEASAR I have not finished fully testing it on real neurons, but if you wanted to contribute there that would be useful!

Thank you! I will check it out some time next week :date:

This repo has only numpy and scipy as dependancies.

Amazing. :man_dancing: