Intro & user manual#
Yet another graph library?#
It is not ;)
This library is based on existing graph libraries (such as graph-tool, igraph, networkx, and possibly soon SNAP) and acts as a convenient interface to build various networks from efficient and verified algorithms. Most importantly, it provides a series of analysis functions that are guaranteed to provide the same results with all backends, enabling fully portable codes (see Consistent tools for graph analysis).
Moreover, it also acts as an interface between those graph libraries and the NEST and DeNSE simulators.
For users that are in a hurry, you can go directly to the Tutorial section. For more specific and detailed examples, several topics are then detailed separately in the following pages:
This library provides many tools which will (or not) be loaded on startup depending on the python packages available on your computer. The default behaviour of those tools is set in the ~/.nngt/nngt.conf file (see Configuration). Moreover, to see all potential messages related to the import of those tools, you can use the logging function of NNGT, either by setting the log_level value to INFO, or by setting log_to_file to True, and having a look at the log file in ~/.nngt/log/.
The graph objects#
Neural networks are described by four graph classes which contain a graph
object from the chosen graph library (e.g.
Graph: base for simple topological graphs with no spatial structure, nor biological properties
SpatialGraph: subclass for spatial graphs without biological properties
Network: subclass for topological graphs with biological properties (to interact with NEST)
SpatialNetwork: subclass with spatial and biological properties (to interact with NEST)
Using these objects, the user can access to the topological structure of the network (for neuroscience, this includes the connections’ type – inhibitory or excitatory – and its synaptic weight, which is always positive)
Nodes/neurons are defined by a unique index which can be used to access their properties and those of the connections between them.
The graph objects can have other attributes, such as:
SpatialNetwork, describes the spatial delimitations of the nodes’ environment (e.g. many in vitro culture of neurons are contained in circular dishes),
structuredivides the graph into groups and can facilitate graph generation and analysis,
Network, contains informations on the various groups of neurons that exist in the network (for instance inhibitory and excitatory neurons can be grouped together), and is the updated version of
structurefor neuroscientific projects.
Several classical graphs are efficiently implemented and the generation procedures are detailed in the documentation.
nngt.simulationdirectly in python causes a
ValueError: module object substituted in sys.modules during a lazy loadwhich I don’t know how to avoided… use
from nngt.geospatial/simulation import whatever_you_wantor
import nngt.geospatial/simulation as ng/nsinstead.
See the issue trackers on Codeberg or GitHub for up-to-date lists.