synapse_reviewer

Neuroglancer project to allow a reviewer to classify putative synapses as real or false-positive

This project runs Neuroglancer with Python callbacks to review synapses. The user runs synapse-reviewer and is presented with synapses in Neuroglancer to rate as true positives or false negatives. There are several keyboard commands within Neuroglancer:

• shift+Y: yes, it is a synapse
• shift+N: no, it is not a synapse
• shift+S: skip to next synapse
• shift+B: go back to the previous synapse
• shift+R: refresh the display

Installation

To install the Python dependencies

> git clone https://github.com/google/neuroglancer
> cd neuroglancer/python
> pip install --editable .
> git clone https://github.com/microns-ariadne/synapse_reviewer
> cd synapse_reviewer
> pip install --editable .

In addition, you have to install the Python dev server on your machine. Follow steps 1-3 here: https://github.com/google/neuroglancer#building Then start the webserver using the command, npm run dev-server-python

The Python install should install the synapse-reviewer program into your environment. Running synapse-reviewer will start a webserver serving the Python application.

Command-line arguments.

Invoke the reader like this:

synapse-reviewer
     [--port <PORT>]
     [--static-content-source <SOURCE>]
     [--image-url <IMAGE_URL>]
     [--segmentation-url <SEGMENTATION_URL>]
     [--synapses <SYNAPSE_CONNECTIONS_FILE>]
     --output <RESULTS_FILE>

where:

• PORT is the port that the Python webserver runs on (default is 8081).
• SOURCE is the URL of the static content webserver. (default is http://localhost:8080 which would be the webserver that is started on your local machine with the npm run dev-server-python command).
• IMAGE_URL is the Neuroglancer / Butterfly / Dojo url of your image datasource. The default is ndstore://https://dojo.rc.fas.harvard.edu/R0::2017_08_23::88_88_14/1_1_1_raw which is the URL within Harvard of our 100um cube. If you are within Harvard, you can ask one of the Dojo / Butterfly admins to host your image datasource there. If you're outside of Harvard, you can make your own arrangements to host a datasource serving one of the Neuroglancer-supported formats.
• SEGMENTATION_URL is the Neuroglancer / Butterfly / Dojo url of your segmentation datasource. By default, this is a matching segmentation to the R0 image datasource. See IMAGE_URL above for setting up your own.
• SYNAPSE_CONNECTIONS_FILE is the path to a JSON file containing the segment IDs of the pre- and post- synaptic neurons and the X, Y and Z locations of the synapse in voxel coordinates. See https://github.com/microns-ariadne/pipeline_engine#synapse-connectionsjson-file-format for a description of this file format. The segment IDs should match those in the volume named by the SEGMENTATION_URL.
• OUTPUT This is the filename that will hold the results of the user's classification. See below for file format. Results are appended to the end of the output file, allowing multiple users to classify different sections or allowing a single user to edit the same file across sessions.

Format of output file.

The output file is a CSV file with the following fields:

• synapse_id The index of the synapse within the SYNAPSE_CONNECTIONS_FILE
• neuron_1 The segment ID of the presynaptic neuron.
• neuron_2 The segment ID of the postsynaptic neuron.
• x The voxel coordinate of the synapse in the X direction
• y The voxel coordinate of the synapse in the Y direction
• z The voxel coordinate of the synapse in the Z direction
• class Either "yes" or "no" depending on whether the user accepted or rejected the synapse.