New release of NeuroML: v2beta4

[Padraig Gleeson]

Dear colleagues,

We are happy to announce a new stable release (version 2, beta 4) of the NeuroML language for model specification in computational neuroscience.

For links to the schemas, specifications, libraries, examples and documentation related to this release please see: http://www.neuroml.org/getneuroml

A summary of the changes in this version follows:

• Renamed the main Schema from NeuroML_v2beta3.xsd to NeuroML_v2beta4.xsd. Changes outlined below are reflected in the new schema

Cell types

• Added <pinskyRinzelCA3Cell>, based on Pinsky and Rinzel 1994. See here for more. Thanks to Justas Birgolas for this.

• Added <izhikevich2007Cell>. Version of Izhikevich cell from 2007 book has been added. Main advantage is explicit capacitance, allowing dimensional currents to be applied. See here for more.

• Added <fitzHughNagumo1969Cell> Version of FitzHugh Nagumo cell based on 1969 model. See https://github.com/OpenSourceBrain/FitzHugh-Nagumo. Thanks to Boris Marin.

• Cells with 2 independent pools of Ca2+. The <cell2CaPools> has been added for cells with 2 independent pools of Ca2+. This may be required where some Ca channels contribute to changes in internal Ca2+ and some don't (just pass charge). See here for example. Thanks to Rokas Stanislovas.

Synapses

• Gap junctions and analog synapses. Added support for gap junctions (example) and analog synapses (example). Supported in jLEMS and NEURON mapping via jNeuroML.

• New synapse types <alphaSynapse> and <expThreeSynapse>. Added synapse types <alphaSynapse> (rise time = decay time) and <expThreeSynapse> (1 exponential rise time, 2 decay times). See here.

• Improved recording from multiple synapses on multicompartmental cells. See here for an example of recording/saving of different variables on multiple synapses on a multicompartmental cell.

Ion channels/conductances

• Fractional conductances from sub gates in channels. Added <gateFractional> which allows multiple children <subGate>, each of which has a fractional conductance. See here and here. Thanks to Boris Marin.

• Gate with instantaneous opening. Added <gateHHInstantaneous> with only <steadyState> child, i.e. no time dependence of opening, just a voltage dependent steady state.

• Improved support for kinetic scheme based channels. Channels based on kinetic schemes (using <ionChannelKS> and <gateKS>) have much better support in jLEMS and NEURON via jNeuroML, see here and here.

• Alternative GHK channel density. Added a second mechanism for specifying channel densities which lead to currents based on the Goldman Hodgkin Katz current, <channelDensityGHK2>. See here and here for examples. Thanks to András Ecker.

Network

• Connections with weights and delays. Better support in jNeuroML and NEURON for <connectionWD>. See here.

Input/output

• Additional spiking/current inputs. New types of inputs to apply to cells, including <poissonFiringSynapse>, <transientPoissonFiringSynapse>, <timedSynapticInput> and <compoundInput>. See here for examples. Thanks to Rokas Stanislovas.

• Saving of spike times. Added support for specifying in LEMS simulation file that spike times should be saved (as opposed to full membrane potential trace). See here for more. Thanks to Finn Krewer.

Testing

• OMV tests on core examples. Added tests on core LEMS examples using OMV (Open Source Brain Model Validation framework). This can be used to run the examples with jNeuroML and other simulators and ensure correct spike times, etc. See the LEMSexamples/test directory. The output from these tests on 16 different simulator configurations can be seen here. Thanks to Boris Marin.

• OMV tests on OSB models. There are ~30 different projects on OSB using NeuroML 2 which are being tested using the OMV framework. See this page for an overview of these.

Tool support

• libNeuroML updated. libNeuroML, the Python API for reading/writing NeuroML2 has been regenerated from the latest Schema. There is also better support for Python 3.

• pyNeuroML is a well tested alternative to jNeuroML. pyNeuroML is a Python module which can be installed with pip install pyNeuroML and can be used to access most of the functionality of jNeuroML. Thanks to Rick Gerkin for testing/updating.

Documentation

• Consolidated web page for documentation. Direct links to all schemas, papers, libraries, tools, examples, contact details are now available on one page: https://neuroml.org/getneuroml

Note, only contributors who are not NeuroML Editors are specifically pointed out above.

Regards,
The NeuroML Editorial Board
Upi Bhalla
Robert Cannon
Sharon Crook
Andrew Davison
Padraig Gleeson

-----------------------------------------------------
Padraig Gleeson
Room 321, Anatomy Building
Department of Neuroscience, Physiology&  Pharmacology
University College London
Gower Street
London WC1E 6BT
United Kingdom

+44 207 679 3214
p.gl...@ucl.ac.uk
-----------------------------------------------------