The NGV anatomy model is reconstructed using the NGV circuit building framework, written in Python, using a component-based design. A component is defined as a standalone processing step with strictly-defined inputs and outputs. The execution of a component may depend on the output files of more than one other components. This relationship between components results in an acyclic graph of dependencies (DAG), which is managed by the Snakemake workflow engine (Köster and Rahmann, 2012).
There are three main datasets in the NGV digital reconstruction: the neuronal, glial, and vascular datasets linked together with pairwise connectivities. The neuronal circuit was generated using the pipeline published in Markram et al., 2015. Protoplasmic astrocytes were placed and grown in the space populated by the vasculature and neurons. Their processes reached the vascular surface and linked with neuronal synapses. A total of 14,402 astrocytes occupied the bounding region, co-localized with 88,541 neurons and 1.37 m of vasculature wiring. The central microcircuit consisted of 15,888 neurons, 2,469 astrocytes, and 0.23 m of the vasculature, occupying a volume of 0.2 mm3.