GENESIS: Documentation

Related Documentation:

Introduction

The createprojection command of the gshell allows convenient creation of projections between two populations of multicompartmental neurons. A projection can have one or more connections.

Implementation

The createprojection command is a user-friendly wrapper around the low-level command volumeconnect. Functionally these commands are the same and the volumeconnect should normally not be used. Their implementation is based on the GENESIS-2 (G-2) commands for the creation of connections (G-2 planarconnect), and setting their weights (G-2 planarweight) and delays (G-2 planardelay). It is instructive to read the overview of these G-2 functions to develop an understanding of their functionality.

Specification

An example invocation of the createprojection command¹ is shown below:

We call the items before the arrow signs (=>) dictionary keys or keys. The invocation distinguishes four main of the parameterization of the command:

• Periphery configuration includes setting of the name of the network (root), the name of the projection that will be created as part of the model (the name key inside the projection key).

  The (probability and random_seed keys

  root => ’/RSNet’,         projection => {                        name => ’/RSNet/projection’,   #                    source => ’../population’, # optional   #                    target => ’../population’, # optional                       },         probability => ’1.0’,         random_seed => ’1212.0’,

• The source key defines the model components that are selected as candidates for outgoing connections. In this case all the neurons in the model component with name /RSNet/population are selected (more precisely all the neurons in a box shaped volume with size 2e10 meters).

  source => {                    context => ’/RSNet/population’,                    include => {                                type => ’box’, # type => ’all’ would remove the need for the line below                                coordinates => [ ’-1e10’, ’-1e10’, ’-1e10’, ’1e10’, ’1e10’, ’1e10’, ],                               },                   },

• The target key defines the model components that are selected as candidates to receive connections. In this case for each neuron found in the source region, an attempt will be made to pair it with any of the neurons found by the include key, except for the neurons that are excluded by the excluded key.

  target => {                    context => ’/RSNet/population’,                    include => {                                type => ’ellipse’,                                coordinates => [ 0, 0, 0, $SEP_X * 1.2, $SEP_Y * 1.2, $SEP_Z * 0.5, ],                               },                    exclude => {                                type => ’box’,                                coordinates => [ - $SEP_X * 0.5, - $SEP_Y * 0.5, - $SEP_Z * 0.5, $SEP_X * 0.5, $SEP_Y * 0.5, $SEP_Z * 0.5, ],                               },                   },

• The verbatim key defines the properties of the individual connections. The pre and post keys specify the names of the pre- and post-synaptic components in the model respectively. Other names are excluded from the pairing process that instantiates the connections.

  synapse => {                     pre => ’spike’,                     post => ’Ex_channel’,                     weight => {   #                            type => ’fixed’, # not sure yet, weight distribution example needed                                value => $syn_weight,                               },                     delay => {   #                           type => ’fixed’, # not sure yet, variable delay example needed                               value => $SEP_X / $cond_vel,   #                           velocity => 0.5, # alternative to fixed delay?                              },                    },

Terminology

• projection: contains a set of connections. Connections can optionally be embedded in connection_groups.
• population: a set of neurons.
• neuron: a neuron model with or without a realistic morphology.
• connection: an attachment between a pre- and post-synaptic component in a model. A connection can hold a weight and a delay.