understanding-datasets.md

Working with Nemanode datasets

Here is a JSON snippet from one of the Nemanode datasets:

[
  {
    "ids": [
      278,
      278,
      280
    ],
    "post": "BWM-DL02",
    "post_tid": [
      7063254,
      7063254,
      7063108
    ],
    "pre": "RMHR",
    "pre_tid": [
      7059560,
      7059560,
      7059620
    ],
    "syn": [
      1,
      1,
      1
    ],
    "typ": 0,
    "datasetType": "head",
    "datasetId": "TEM_L3"
  },
  ...
]

Each dataset consists of a list of connections.

Each connection will have the following structure:

  {
    "ids": [
      278,
      278,
      280
    ],
    "post": "RIML",
    "post_tid": [
      7063254,
      7063254,
      7063108
    ],
    "pre": "AWAL",
    "pre_tid": [
      7059560,
      7059560,
      7059620
    ],
    "syn": [
      1,
      1,
      1
    ],
    "typ": 0,
    "datasetType": "head",
    "datasetId": "TEM_L3"
  }

Understanding the connection data fields

Each connection has a set fields describing the connection:

• pre: name of the post-synaptic cell (string) e.g. AWAL
• post: name of the post-synaptic cell (string) e.g. RIML
• typ: the type of the connection
  • 0: means that the connection is a chemical synapse
  • 2: means that the connection is a electrical synapse (gap junction)
• datasetType: the location of the connection, one of tail, head complete
• datasetId: the id of the dataset that the connection is apart of

Each connection also has a set of fields that describe each of the synapses that it represents

• ids: a list of synapse ids of size N. syn[0] corresponds to the first synapse, syn[1] corresponse to the second synapse, etc.
• syn: the number of synapses encoded as a list of numbers. e.g. [1, 1, 1] means there is 3 synapses in the connection
• post_tid: the ids of the post synaptic tree nodes of the synapses (used for the 3d neuron trajectory visualization)
  • Note: These are not present in the legacy datasets
• pre_tid: the ids of the pre synaptic tree nodes of the synapses (used for the 3d neuron trajectory visualization)
  • Note: These are not present in the legacy datasets