Neuroarch: A Graph-Based Platform for Constructing and Querying Models of the Fruit Fly Brain Architecture
-
1
Columbia University, United States
The fruit fly Drosophila melanogaster is an excellent model organism for reverse engineering information processing in biological brains due to its capacity for complex nonreactive behavior mediated by a brain containing a relatively small number of computational components. An increasingly powerful toolbox of genetic and electrophysiological techniques enables researchers to experimentally relate the fly’s behavior to its brain structure. Efforts to fully map the fly’s connectome have identified fewer than 50 distinct functional units in its brain, most of which are characterized by unique populations of local neurons [1, 2, 3]. The increasing volume of available biological data regarding the structure of these local processing units (LPUs) and the connectivity tracts between them must be leveraged in the design of in silico fly brain models.
Successfully modeling the fly brain requires integration and execution of LPU and tract models developed independently by multiple researchers [4, 5]. Software that enables such integration must also afford researchers the freedom to specify the parameters of LPU models that may employ disparate internal designs. Ongoing efforts to obtain increasingly accurate data regarding the fly’s brain structure also demand that such software be sufficiently flexible to facilitate revision of the models in a straightforward manner even if completely new components must be added. Existing tools for structured specification of neural models [6] provide limited means for querying model data required by emulation engines responsible for efficient model execution. Moreover, changes to an LPU model’s internal design may complicate its reimplementation by requiring that support for accessing new design elements be retrofitted into representations of the existing model.
To address the above concerns, we have developed a Python package called Neuroarch for representation and storage of LPU-based models of the fly brain. It provides researchers and software applications with a common interface for defining, querying, and manipulating integrated model data. Neuroarch’s representation of the fly’s brain distinguishes between the connectivity architecture linking its LPUs and the design of the individual LPUs identified in [3]; the former is modeled as graphs of communication ports exposed by each LPU and the connections between them while the latter comprises graphs of the internal modeling elements required by specific LPU models. Neuroarch’s flexibility stems from its storage of all modeling elements comprised by the connectivity architecture and different LPU designs (including synaptic model instances) as graph nodes; edges are exclusively used to represent relation- ships between modeling elements. Neuroarch stores all model data in a graph database to accelerate those queries typically performed to determine which modeling elements must be updated simultaneously during model execution.
Both LPU design and inter-LPU connectivity model data is exclusively accessed through Neuroarch’s object-relational mapping (ORM) of modeling elements to their internal database representations. A key feature of this ORM is its support for multimodal views of query results over stored model data that may be passed as operands to other graph operations. This enables model data to be accessed or modified either as a subgraph (to facilitate graph-based queries) or a tensor (to facilitate tabular or relational queries). Neuroarch’s I/O layer extends and invokes the ORM to load model data expressed in several specification formats such as CSV, GEXF, and NeuroML. Models may also be constructed from basic circuit motifs (specified either as small graphs or tensors) using Neuroarch’s graph composition operators.
We have used Neuroarch to drive fly brain emulations of a prototype multisensory coincidence detection system that integrates 4 independently developed LPUs in the fly vision and olfactory systems executed using the Neurokernel framework [4, 7]; model data for some of these LPUs is explicitly specified, while the remaining LPUs are constructed using Neuroarch’s graph operators to compose canonical circuits identified in those LPUs [8].
Acknowledgements
The research reported here was supported by AFOSR under grant #FA9550-12-1-0232.
References
[1] Chiang, et al. Three-dimensional reconstruction of brain-wide
wiring networks in Drosophila at single-cell resolution. Current Biology,
21(1):1-11, January 2011.
[2] Lin, et al. A comprehensive wiring diagram of the protocerebral bridge
for visual information processing in the drosophila brain. Cell Reports,
3(5):1739-1753, May 2013.
[3] Ito, et al. A systematic nomenclature for the insect brain. Neuron, 81(4):755-765,
February 2014.
[4] Lev E. Givon and Aurel A. Lazar. Neurokernel: An open scalable architecture
for emulation and validation of drosophila brain models on multiple gpus.
Neurokernel Request for Comments, Neurokernel RFC #1, February 2014.
[5] Aurel A. Lazar, Nikul H. Ukani, and Yiyin Zhou. The cartridge: A canonical
neural circuit abstraction of the lamina neuropil - construction and composition
rules. Neurokernel Request for Comments, Neurokernel RFC #2, January 2014.
[6] Gleeson, et al. NeuroML: a language for describing data driven models of neurons and networks with a high degree of biological detail. PLoS Comput Biol, 6(6):e1000815, June
2010.
[7] L. E. Givon and A. A. Lazar. Neurokernel: An open source platform for emulating
the fruit fly brain. 2014, submitted for publication.
[8] A. A. Lazar, W. Li, N. H. Ukani, C.-H. Yeh, and Y. Zhou. Neural circuit abstractions
in the fruit fly brain. In Society for Neuroscience Abstracts, Nov 2013.
Keywords:
Drosophila,
local processing units,
integrated data models,
graph databases,
object-relational mapping
Conference:
Neuroinformatics 2014, Leiden, Netherlands, 25 Aug - 27 Aug, 2014.
Presentation Type:
Poster, not to be considered for oral presentation
Topic:
Large-scale modeling
Citation:
Givon
LE,
Lazar
AA and
Ukani
NH
(2014). Neuroarch: A Graph-Based Platform for Constructing and Querying Models of the Fruit Fly Brain Architecture.
Front. Neuroinform.
Conference Abstract:
Neuroinformatics 2014.
doi: 10.3389/conf.fninf.2014.18.00042
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
04 Apr 2014;
Published Online:
04 Jun 2014.
*
Correspondence:
Prof. Aurel A Lazar, Columbia University, New York, United States, aurel@ee.columbia.edu