[neuroConstruct] NeuroML version 2.0 Milestone: v2alpha

P Gleeson p.gleeson at ucl.ac.uk
Wed Dec 15 11:21:07 GMT 2010

Hi all,

A number of activities have been taking place towards NeuroML version 2. 
These have been mentioned in some previous mails and have been in 
development on the Sourceforge SVN repository, but the various threads 
have reached a point where a draft schema for version 2 (NeuroML 
v2alpha) can be used for testing in applications which import/export 
NeuroML 2.

Full details of the ongoing work are available here: 

Some of the things which are possible with the current version of NeuroML 2:

* Full 3D networks containing morphologically detailed cells with 
electrical properties, ion channels and synapses can be expressed in 
NeuroML v2alpha and validated according to the new XML Schema (i.e. 
roughly the same scope as version 1.x Level 3)

* Abstract cell models including the Izhikevich cell and adapting 
exponential integrate & fire can be specified using Robert's LEMS*, and 
instantiated in NeuroML 2 files. These can be executed with the LEMS 
interpreter or NEURON code can be generated.

* Examples of synapse models including single & double exponential, 
voltage dependent (NMDA-R) and short term plasticity models are 
available, as are HH and kinetic scheme ion channel models. These can be 
extended using the LEMS framework and executed with the interpreter.

* An initial export of models specified in NeuroML 2 (and many other 
models in LEMS) to SBML is available.

* Various aspects of NeuroML 2 development have benefited from 
discussions in the INCF Multiscale Modelling Program, and  files in the 
emerging NineML language can be generated from many of the NeuroML 
2/LEMS examples.

Please look at the provided examples, try out the available tools, try 
building new model components using LEMS (which can subsequently be 
included as core NeuroML 2 types) and give your comments on the main 
NeuroML mailing list. Hopefully we can get some useful feedback on the 
currently proposed structure for NeuroML 2 before the 3rd NeuroML 
development workshop in London at the end of March next year (more 
details to follow).

The NeuroML Team

* To summarise the relation between NeuroML 2 and LEMS (Low Entropy 
Model Specification, http://www.neuroml.org/lems): LEMS can be used 
independently as a language for specifying reusable model types, and is 
not really neuroscience specific. NeuroML 2 XML files (very neuroscience 
specific...) can be used without interaction with LEMS, but most of the 
core NeuroML types will be specified in LEMS and applications can 
benefit greatly from the machine readable specification of model 
component behaviour.

More information about the neuroConstruct mailing list