Boris Marin

LEMS: a language for expressing complex biological models in concise and hierarchical form and its use in underpinning NeuroML 2.

Computational models are increasingly important for studying complex neurophysiological systems. As scientific tools, it is essential that such models can be reproduced and critically evaluated by a range of scientists. However, published models are currently implemented using a diverse set of modeling approaches, simulation tools, and computer languages making them inaccessible and difficult to reproduce. Models also typically contain concepts that are tightly linked to domain-specific simulators, or depend on knowledge that is described exclusively in text-based documentation. To address these issues we have developed a compact, hierarchical, XML-based language called LEMS (Low Entropy Model Specification), that can define the structure and dynamics of a wide range of biological models in a fully machine readable format. We describe how LEMS underpins the latest version of NeuroML and show that this framework can define models of ion channels, synapses, neurons and networks. Unit handling, often a source of error when reusing models, is built into the core of the language by specifying physical quantities in models in terms of the base dimensions. We show how LEMS, together with the open source Java and Python based libraries we have developed, facilitates the generation of scripts for multiple neuronal simulators and provides a route for simulator free code generation. We establish that LEMS can be used to define models from systems biology and map them to neuroscience-domain specific simulators, enabling models to be shared between these traditionally separate disciplines. LEMS and NeuroML 2 provide a new, comprehensive framework for defining computational models of neuronal and other biological systems in a machine readable format, making them more reproducible and increasing the transparency and accessibility of their underlying structure and properties.

Open Source Brain: a collaborative resource for visualizing, analyzing, simulating and developing standardized models of neurons and circuits

Computational models are powerful tools for investigating brain function in health and disease. However, biologically detailed neuronal and circuit models are complex and implemented in a range of specialized languages, making them inaccessible and opaque to many neuroscientists. This has limited critical evaluation of models by the scientific community and impeded their refinement and widespread adoption. To address this, we have combined advances in standardizing models, open source software development and web technologies to develop Open Source Brain, a platform for visualizing, simulating, disseminating and collaboratively developing standardized models of neurons and circuits from a range of brain regions. Model structure and parameters can be visualized and their dynamical properties explored through browser-controlled simulations, without writing code. Open Source Brain makes neural models transparent and accessible and facilitates testing, critical evaluation and refinement, thereby helping to improve the accuracy and reproducibility of models, and their dissemination to the wider community.

Geppetto: a reusable modular open platform for exploring neuroscience data and models

Geppetto is an open-source platform that provides generic middleware infrastructure for building both online and desktop tools for visualizing neuroscience models and data and managing simulations. Geppetto underpins a number of neuroscience applications, including Open Source Brain (OSB), Virtual Fly Brain (VFB), NEURON-UI and NetPyNE-UI. OSB is used by researchers to create and visualize computational neuroscience models described in NeuroML and simulate them through the browser. VFB is the reference hub for Drosophila melanogaster neural anatomy and imaging data including neuropil, segmented neurons, microscopy stacks and gene expression pattern data. Geppetto is also being used to build a new user interface for NEURON, a widely used neuronal simulation environment, and for NetPyNE, a Python package for network modelling using NEURON. Geppetto defines domain agnostic abstractions used by all these applications to represent their models and data and offers a set of modules and components to integrate, visualize and control simulations in a highly accessible way. The platform comprises a backend which can connect to external data sources, model repositories and simulators together with a highly customizable frontend. This article is part of a discussion meeting issue ‘Connectome to behaviour: modelling C. elegans at cellular resolution’.

Development and application of a ray-based model of light propagation through a spherical acousto-optic lens

A spherical acousto-optic lens (AOL) consists of four acousto-optic deflectors (AODs) that can rapidly and precisely control the focal position of an optical beam in 3D space. Development and application of AOLs has increased the speed at which 3D random access point measurements can be performed with a two-photon microscope. This has been particularly useful for measuring brain activity with fluorescent reporter dyes because neuronal signalling is rapid and sparsely distributed in 3D space. However, a theoretical description of light propagation through AOLs has lagged behind their development, resulting in only a handful of simplified principles to guide AOL design and optimization. To address this we have developed a ray-based computer model of an AOL incorporating acousto-optic diffraction and refraction by anisotropic media. We extended an existing model of a single AOD with constant drive frequency to model a spherical AOL: four AODs in series driven with linear chirps. AOL model predictions of the relationship between optical transmission efficiency and acoustic drive frequency including second order diffraction effects closely matched experimental measurements from a 3D two-photon AOL microscope. Moreover, exploration of different AOL drive configurations identified a new simple rule for maximizing the field of view of our compact AOL design. By providing a theoretical basis for understanding optical transmission through spherical AOLs, our open source model is likely to be useful for comparing and improving different AOL designs, as well as identifying the acoustic drive configurations that provide the best transmission performance over the 3D focal region.

Tilted excitation implies odd periodic resonances

Our aim is to unveil how resonances of parametric systems are affected when symmetry is broken. We showed numerically and experimentally that odd resonances indeed come about when the pendulum is excited along a tilted direction. Applying the Melnikov subharmonic function, we not only determined analytically the loci of saddle-node bifurcations delimiting resonance regions in parameter space but also explained these observations by demonstrating that, under the Melnikov method point of view, odd resonances arise due to an extra torque that appears in the asymmetric case.

Noise, transient dynamics, and the generation of realistic interspike interval variation in square-wave burster neurons.

First return maps of interspike intervals for biological neurons that generate repetitive bursts of impulses can display stereotyped structures (neuronal signatures). Such structures have been linked to the possibility of multicoding and multifunctionality in neural networks that produce and control rhythmical motor patterns. In some cases, isolating the neurons from their synaptic network reveals irregular, complex signatures that have been regarded as evidence of intrinsic, chaotic behavior. We show that incorporation of dynamical noise into minimal neuron models of square-wave bursting (either conductance-based or abstract) produces signatures akin to those observed in biological examples, without the need for fine tuning of parameters or ad hoc constructions for inducing chaotic activity. The form of the stochastic term is not strongly constrained and can approximate several possible sources of noise, e.g., random channel gating or synaptic bombardment. The cornerstone of this signature generation mechanism is the rich, transient, but deterministic dynamics inherent in the square-wave (saddle-node and homoclinic) mode of neuronal bursting. We show that noise causes the dynamics to populate a complex transient scaffolding or skeleton in state space, even for models that (without added noise) generate only periodic activity (whether in bursting or tonic spiking mode).

Dynamics of a parametrically excited simple pendulum

The dynamics of a parametric simple pendulum submitted to an arbitrary angle of excitation ϕ was investigated experimentally by simulations and analytically. Analytical calculations for the loci of saddle-node bifurcations corresponding to the creation of resonant orbits were performed by applying Melnikovs method. However, this powerful perturbative method cannot be used to predict the existence of odd resonances for a vertical excitation within first order corrections. Yet, we showed that period-3 resonances indeed exist in such a configuration. Two degenerate attractors of different phases, associated with the same loci of saddle-node bifurcations in parameter space, are reported. For tilted excitation, the degeneracy is broken due to an extra torque, which was confirmed by the calculation of two distinct loci of saddle-node bifurcations for each attractor. This behavior persists up to ϕ≈7π/180, and for inclinations larger than this, only one attractor is observed. Bifurcation diagrams were constructed experimentally for ϕ=π/8 to demonstrate the existence of self-excited resonances (periods smaller than three) and hidden oscillations (for periods greater than three).

Automated code generation from LEMS, the general purpose model specification language underpinning NeuroML2

Making models more transparent, reproducible and accessible is a major challenge in computational neuroscience. The wide range of modelling tools and strategies available, while convenient from a pragmatic point of view, often make model reuse harder. Efforts on standardization [1], model sharing [2,3] and practices learned from the open-source community [3-5], on the other hand, are proving increasingly invaluable in improving the transparency and reusability of computational models. NeuroML2, the latest version of a model description language for computational neuroscience, is built on LEMS (Low Entropy Model Specification, [6]) a general purpose model specification language. LEMS has a highly structured, hierarchical format, rooted in physical principles, which allows the definition of physical systems including arbitrary networks of nodes modeled as hybrid dynamical systems. There are currently two LEMS interpreters, jLEMS [5] and PyLEMS [6], implemented in Java and Python respectively. As reference implementations, numerical performance and accuracy are not their primary goals. In order to simulate NeuroML2/LEMS models in an accurate, robust and efficient way it is important to be able to perform the numerical integration using well tested scientific simulators, both neuroscience specific, e.g. NEURON, and general purpose, e.g. XPPAUT, MATLAB. Such simulators provide a robust numerical infrastructure important for modelling in general, such as high-order ODE integration and accurate event detection. We have developed a set of template-based code generators for LEMS whose goal is to transform a LEMS model into a number of representations compatible with different simulators, using back-end specific templates to produce native code. The translation of the models happens through an intermediate format, developed for this purpose, named dLEMS (distilled LEMS). The dLEMS abstraction is ODE centric for convenience of interoperability with the different target languages. It represents a flattened (e.g. hierarchical channel descriptions are reduced to the core ODEs) and dimensionless version of the LEMS model. The advantage of this approach, compared for instance to a compiler based strategy, is that users willing to provide support for a target simulator can more easily work with dLEMS, as it closely reflects the ODE representations used in general purpose simulators. Users will simply have to create new templates for their target simulator or modify existing ones. Our initial results show that the template-based code-generation approach is feasible, producing runnable simulations in a variety of platforms for single cell models – in particular, templates are available for the following platforms: C code generation (using the LLNL Sundials library for numerical integration), XPPAUT, MATLAB and Modelica. All tools we have developed for this are open source and freely available. We are presently working at extending the dLEMS specification of connections between elements, creating a simplified format for template based mapping to neuronal simulators.

Temporal structure of bursting patterns as representation of input history

From Eighteenth Annual Computational Neuroscience Meeting: CNS*2009 Berlin, Germany. 18–23 July 2009.