Neuclease

Tools for computing interactive "cleaves" of agglomerated neuron fragments from a DVID server.

Install

conda install -c flyem-forge -c conda-forge neuclease

Server Launch Example

neuclease_cleave_server \
    --port 5555 \
    --merge-table /path/to/merge-table.npy \
    --primary-dvid-server emdata3:8900 \
    --primary-uuid 017a \
    --primary-labelmap-instance segmentation \
    --log-dir /path/to/logs

Example Client Request

import requests

data = { "user": "bergs",
         "body-id": 673509195,
         "port": 8900,
         "seeds": {"1": [675222237], "2": [1266560684], "3": [1142805921]},
         "server": "emdata3.int.janelia.org", # DVID server
         "uuid": "017a",
         "segmentation-instance": "segmentation",
         "mesh-instance": "segmentation_meshes_tars" }

r = requests.post('http://bergs-ws1.int.janelia.org:5556/compute-cleave', json=data)
print(r.json())

# { "assignments": {
#      "1": [675222237, 12345, ...],
#      "2": [1266560684, 23456, ...],
#      "3": [1142805921, ....] },
#   ...
# }

Optional dependency

One step in the algorithm can be sped up slightly if you install graph-tool:

conda install -c flyem-forge graph-tool

(Otherwise, networkx is used. It doesn't matter much, except for very large cleaves.)