neurarc-py

Python reader/writer for the ARC binary connectome format.

Install

pip install neurarc

# with optional extras
pip install neurarc[scipy]      # scipy sparse matrix export
pip install neurarc[networkx]   # networkx graph export
pip install neurarc[sonata]     # SONATA HDF5 import
pip install neurarc[all]        # everything

Quick Start

from neurarc import load, save, from_csv, from_json, validate

# import from CSV
arc = from_csv("edges.csv")
save("network.arc", arc)

# import from JSON
arc = from_json("edges.json")
save("network.arc", arc)

# load and validate
arc = load("network.arc")
errors = validate(arc)

CLI

# import from CSV/JSON/SONATA
neurarc import edges.csv -o network.arc
neurarc import edges.json -o network.arc
neurarc import nodes.h5 -o network.arc -f sonata --edges-file edges.h5

# export to CSV
neurarc export network.arc -o edges.csv -f csv

# inspect
neurarc info network.arc
neurarc validate network.arc
neurarc dump-header network.arc
neurarc dump-tensors network.arc --values

# convert between formats
neurarc convert network.arc -f csr -o network_csr.arc

Python API

import neurarc as arc

# import
arc_obj = arc.from_csv("edges.csv", neuron_count=1000)
arc_obj = arc.from_json("edges.json")
arc_obj = arc.from_sonata("nodes.h5", "edges.h5")

# save/load
arc.save("network.arc", arc_obj)
arc_obj = arc.load("network.arc")

# convert
csr = arc.to_csr(arc_obj)
back = arc.to_edge_list(csr)

# export
arc.to_csv("output.csv", arc_obj)
scipy_mat = arc.to_scipy_csr(arc_obj)
nx_graph = arc.to_networkx(arc_obj)

Development

pip install -e ".[dev]"
pytest