bdpedigo / neuropull

A (soon to be) lightweight Python package for accessing single-cell connectome networks with metadata.

neuropull

A lightweight tool for pulling connectome networks and metadata

• Documentation: https://neurodata.github.io/neuropull
• GitHub: https://github.com/neurodata/neuropull
• PyPI: https://pypi.org/project/neuropull/
• Free software: MIT

Features

• TODO

Credits

This package was created with Cookiecutter and the waynerv/cookiecutter-pypackage project template.

neuropull's Issues

Write generic network loader function

write data wrangling classes

Desiderata

ArraySubgraph

(?)

ArrayGraph

• union of one graph with another
• intersect of one graph with another
• largest connected component
• add two graphs (only if indexed the same way)
• reindex
• reindex_like
• query (subselect) by node info
• groupby
  • induced subgraph only?
• sort_values
• len
• repr
• convert to networkx (appropriately for symmetry class)

MultiArrayGraph

• initialize from union of graphs on different node sets
• reindex
• reindex_like
• query
• groupby
• sort_values
• len
• repr
• intersect lcc
• union lcc
• convert to networkx with different weight attributes

SparseArrayGraph

(?)

SparseMultiArrayGraph

(?)

how to store multiplex networks

column for each type in the edgelist
cons:

• storage (more 0s to keep track of)

pros:

• readability
• still a valid edgelist, easy to read in

row for each type of edge
basically the inverse of the above

add a gallery of the available datasets

connectomes to add as first pass

• c. elegans male
• c. elegans herm
• planynerius
• witvilet c. elegans time series
• p. pacificus pharynges
• d. larva subset

Write spec for network storage

Edgelist

• .csv file
• source, target, weight header. weight can be omitted if unweighted
• networks assumed directed by defaults, some way to specify if not?
• extra per-edge info can be stored as well, always after weight

Node metadata

• Need to codify field types
  • E.g., categorical, numeric, etc
  • Want to have some way of encoding symmetries: e.g. this neuron in this segment etc etc

Graph metadata

• JSON?

add citation info for the available datasets