Brain Atlas Toolkit

This package is intended as a toolkit for manipulating hierarchical brain atlases. For example, the Allen Mouse Brain Atlas has a parent-child ontology of this structure:

{
 "id": 997,
 "acronym": "root",
 "name": "root",
 "graph_order": 0,
 "parent_structure_id": null,
 "children": [
  {
   "id": 8,
   "acronym": "grey",
   "name": "Basic cell groups and regions",
   "graph_order": 1,
   "parent_structure_id": 997,
   "children": [
    {
     "id": 567,
     "acronym": "CH",
     "name": "Cerebrum",
     "graph_order": 2,
     "parent_structure_id": 8,
     "children": [
     ...

Requirements

• python>=3.7
• A system-wide installation of graphviz: https://www.graphviz.org/ if you are going to use any of the visualization tools in this package.

Examples

Load a custom brain atlas from JSON file

from brain_atlas_toolkit import graph_tools
import json

json_file = "allen_ontology.json"
with open(json_file,'r') as infile:
	ontology_dict = json.load(infile)

Note that this JSON file must have the structure of the example ontology shown above. The minimal set of keys in each element are:

• id
• name
• parent_structure_id

Initialize ontology graph

ontology_graph = graph_tools.Graph(ontology_dict)

Get all progeny (a.k.a. descendents or subregions) of a region of interest returned in a flattened list

ontology_graph.get_progeny('Somatomotor areas')

which returns:

['Somatomotor areas, Layer 1', 'Somatomotor areas, Layer 2/3', 'Somatomotor areas, Layer 5', 'Somatomotor areas, Layer 6a', 'Somatomotor areas, Layer 6b', 'Primary motor area', 'Primary motor area, Layer 1', 'Primary motor area, Layer 2/3', 'Primary motor area, Layer 5', 'Primary motor area, Layer 6a', 'Primary motor area, Layer 6b', 'Secondary motor area', 'Secondary motor area, layer 1', 'Secondary motor area, layer 2/3', 'Secondary motor area, layer 5', 'Secondary motor area, layer 6a', 'Secondary motor area, layer 6b']

Get the parent name of a region of interest

ontology_graph.get_parent('Somatomotor areas')

which returns:

Isocortex

Get the integer id of a region of interest

ontology_graph.get_id('Somatomotor areas')

which returns:

500

Get the acronym name of a region of interest

ontology_graph.get_parent('Somatomotor areas')

which returns:

MO

Print a branch of the ontology

ontology_graph.print_branch('Somatomotor areas')

which returns

 0 Somatomotor areas
	 1 Somatomotor areas, Layer 1
	 1 Somatomotor areas, Layer 2/3
	 1 Somatomotor areas, Layer 5
	 1 Somatomotor areas, Layer 6a
	 1 Somatomotor areas, Layer 6b
	 1 Primary motor area
		 2 Primary motor area, Layer 1
		 2 Primary motor area, Layer 2/3
		 2 Primary motor area, Layer 5
		 2 Primary motor area, Layer 6a
		 2 Primary motor area, Layer 6b
	 1 Secondary motor area
		 2 Secondary motor area, layer 1
		 2 Secondary motor area, layer 2/3
		 2 Secondary motor area, layer 5
		 2 Secondary motor area, layer 6a
		 2 Secondary motor area, layer 6b

Print a branch of the ontology and control the depth of the tree

ontology_graph.print_branch('Somatomotor areas',stoplevel=1)

which returns

 0 Somatomotor areas
	 1 Somatomotor areas, Layer 1
	 1 Somatomotor areas, Layer 2/3
	 1 Somatomotor areas, Layer 5
	 1 Somatomotor areas, Layer 6a
	 1 Somatomotor areas, Layer 6b
	 1 Primary motor area
	 1 Secondary motor area

The default stoplevel value is -1, which means print the entire tree to max depth.

Visualize a branch of the ontology

digraph = ontology_graph.visualize_graph('Somatomotor areas',level=2)
digraph.format='png' # control the image type; supports png, pdf and other formats
digraph.view()

The last line will save and then open up an image in your default image viewer application. The image should look like this (click image to view zoomed in version):

For full documentation of the digraph object, see the graphviz Python API documentation: https://graphviz.readthedocs.io/en/stable/