NeuroML

NeuroML is an open, simulator-independent description language and set of schemas for articulating neuronal models and networks. Built to improve sharing, reuse, and reproducibility, NeuroML encodes neuronal morphology, biophysics, synapses, and connectivity in a common, machine-readable format that can be imported by multiple simulation engines.

NeuroML is organized into modules. Morphology descriptions (MorphML) capture neuron geometry, including compartments and branching. Channel and membrane properties (ChannelML) define ionic currents, conductances, kinetics, and passive properties. Synapses are described with SynapseML, while networks describe populations of cells, their connectivity (projections), stimuli, and recordings through the NeuroML Core vocabulary. NeuroML models can be extended with LEMS, a separate language that specifies the dynamical equations and parameter interfaces in a simulator-neutral way, enabling precise translation to target simulators.

LEMS provides a formal basis for representing dynamic behavior and parameters, allowing NeuroML models to be deployed in diverse simulators such as NEURON, NEST, GENESIS, and Brian through translation layers and tools.

Tools and ecosystem: PyNeuroML and JNeuroML offer libraries and command-line tools to validate, convert, and run NeuroML models; NeuroML-DB hosts a repository of NeuroML-encoded models. Community adoption emphasizes reproducibility and cross-platform exchange.

History and scope: NeuroML emerged from a collaborative effort to standardize neuronal model exchange and evolve with evolving simulation technologies. It is maintained by an international community of researchers and developers and remains actively used in computational neuroscience to describe single-neuron models and larger networks.