Data processing documentation

[Hashir Ahmad]

Hi,

Is there any technical documentation or UML workflows for the data processing performed under the Neuroglancer python bridge? The examples seem to be not enough for me to understand the data parsing and processing.

Thank you

[Jeremy Maitin-Shepard]

Unfortunately there is not, I can try to write some but do you have a more specific question?

The basic idea is that Python runs a webserver that serves several purposes:

- hosts the neuroglancer client

- provides a sockjs (similar to websocket) server used to synchronize/control the viewer json state, set up special Python-only configuration options like custom key and mouse bindings, request screenshots, and send credentials (e.g. for accessing Google cloud storage buckets)

- provides additional endpoints serving volumetric data, auto-generated meshes, and skeletons for in-memory python data sources (in python these correspond to LocalVolume and Skeletons piece, in the client these are python:// data sources).

--
You received this message because you are subscribed to the Google Groups "Neuroglancer" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neuroglancer...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/neuroglancer/87b0a779-d7a0-4c34-9b56-ec299cf7aca0n%40googlegroups.com.

[Hashir Ahmad]

I wanted to run a skeleton example for custom data. I already have the meshes precomputed. However, I didn't understand the use of get_skeleton() method under SkeletonSource. It is implemented in the example_skeleton.py but not used anywhere. Unless it is invoked with the instantiation of SkeletonSource. I also opened an issue on the GitHub repo about this. 

Once I know how the skeleton data is parsed, I just want to load the mesh data along with it. 

[Jeremy Maitin-Shepard]

It turns out example_skeleton.py had become broken.  I fixed it.

Currently the Python API can auto-generate and serve meshes for small volumes, but there isn't a mechanism to serve precomputed meshes via the Python API.  That could in principle be added, but you could instead just write your meshes to precomputed format.

To view this discussion on the web visit https://groups.google.com/d/msgid/neuroglancer/92ef2ed1-96c5-4d6e-9626-f2ef9a57515dn%40googlegroups.com.