Tag Archives: Semantic Web

Using Graph Theory in the Brain Sciences

Graph theory has provided a new set of tools for helping us to understand signal processing networks in the brain (see “Other related blog posts” below for some earlier posts on graph theoretic based approaches). In particular, a field known as network theory or complex network theory, which is rooted in graph theory, has been helping to provide insight into the way circuits in the brain are wired and how those circuits contribute to brain function. The new review paper “Dissecting functional connectivity of neuronal microcircuits: experimental and theoretical insights” (published online April 2, 2011 in Trends in Neuroscience) provides a high-level look at using network theory in the brain sciences.

First the authors explain key definitions for those unfamiliar with network theory. They discuss the features of networks that are typically captured in network theoretic equations. These equations are used for analyzing graphs composed of nodes and edges (links). Results of the analyses provide insights into the physical and functional organization of graphs.

The nature of the insights provided by the analysis of a graph depend on the graph’s composition. Nodes may represent any number of brain structures. For example, nodes may represent neurons, brain areas, or dendritic spines. Edges represent the means of communication amongst the structures represented by the nodes. For example, edges may represent chemical synapses, gap junctions, or diffusible molecules.

In network theory some general organizational principles have emerged. For instance, a scale-free network is a graph containing nodes that exhibit a wide range in their number of connections with other nodes. These graphs include rare hub nodes that have an extraordinarily large number of connections with other nodes. Hub nodes have a strong impact on signal processing within a network.

Scale-free networks are found in neuronal circuits in the brain. For example, an important structure for learning and memory known as the hippocampus exhibits scale-free network organization during development. The inhibitory GABAergic neurons in this structure act as hubs that help orchestrate synchronous activity across the network. If these concepts or their application in the brain sciences are interesting and new to you then you may find this review paper to be a good introduction.


Other related blog posts:

New Analysis Methodologies and the Case for Data Sharing in Brain Research

Sex Matters But the Brain is Like Nothing Else

Explosive Change in Network Connectivity and Catastrophic Information Loss

Synthetic Brain Cells and Graph Theory

A Virtual Fly Brain

The Virtual Fly Brain beta website.
Figure 1. The Virtual Fly Brain beta website. Click on Show all Domains to view the controlled vocabulary displayed down the right-hand side of the page.

This past December I wrote “I’d love to be able to use a virtual fly nervous system” (see How the Brain Works, Flies, and the FlyBase Online Data Repository). A week ago David Osumi-Sutherland from the FlyBase Consortium posted a response that “A beta version of Virtual Fly Brain is now live” (see Figure 1 above).

In that previous blog post one of my questions was where the supraesophageal ganglion was located in the fly brain. I got a verbal description but no visual help. In the Virtual Fly Brain the supraesophageal ganglion is listed in the controlled vocabulary under Anatomy Tree and Search along the right-hand side of the page.

Note: The Virtual Fly Brain page may display “Show all Domains” rather than list the anatomy terms. Click on “Show all Domains” to display the anatomy terms down the right-hand side of the page under the “Anatomy Tree and Search” heading.

The supraesophageal ganglion includes the colorized area shown in in the fly brain displayed by the Virtual Fly Brain website.
Figure 2. The supraesophageal ganglion includes the colorized area shown in in the fly brain displayed by the Virtual Fly Brain website.

To highlight the supraesophageal ganglion in the fly brain image, first click on Clear all Selections to remove the terms that are selected by default. Next, click the check box to the left of the supraesophageal ganglion listing in the anatomy tree. The area defined as the supraesophageal ganglion fills with a transparent color shown in the little box to the right of the check box.

The anatomy listing is a display of the Drosophila Brain anatomy ontology, which was developed using OWL2. The ontology provides a means to interlink fly anatomy with innumerable pieces of data. I look forward to exploring more of this promising tool as it develops.


Other related blog posts:

Viewing the Fly Brain Connectome with Brainbow

How the Brain Works, Flies, and the FlyBase Online Data Repository

Making Neuroinformatics Integral to Studying the Brain

The current issue of Science includes the special section Dealing with Data. In it there is a perspective paper on neuroscience and data titled “Challenges and Opportunities in Mining Neuroscience Data” (published February 11, 2011 in Science). The focus of the perspective is on the Human Connectome Project and the Neuroscience Information Framework. These important projects are familiar to those of you who have been following my posts (see “Other related blog posts” below). However, the authors point out a very important fact. Most neuroinformatics resources remain underused by the research community.

The February 14, 2011 cover of Science. The issue includes a special section Dealing with Data.
Figure 1. The February 14, 2011 cover of Science. The issue includes a special section Dealing with Data.

The authors conclude with eight suggestions on how to make neuroinformatics integral to studying the brain:

  • Neuroscientists should share their data and in a form that is easily accessible.
  • Neuroscience databases need to be created, populated, and sustained with adequate support from federal and other funding mechanisms.
  • Databases become more useful as they are more densely populated so adding to existing databases may be preferable to creating new ones.
  • Data consumption will increasingly involve machines first and humans second. Neuroscientists should annotate content using community ontologies and identifiers. Coordinates, atlas, and registration method should be specified when referencing spatial locations.
  • Some types of published data should be reported in standardized table formats that facilitate data mining.
  • Investment needs to occur in interdisciplinary research to develop computational, machine-learning, and visualization methods for synthesizing across spatial and temporal information tiers.
  • Educational strategies from undergraduate through postdoctoral levels are needed to ensure that neuroscientists of the next generation are proficient in data mining and using the data-sharing tools of the future.
  • Cultural changes in the neurosciences are needed to promote widespread participation in this endeavor.

These suggestions, if followed, would certainly move the neuroscience community in the right direction. They seem to assume, however, that human consumption of neuroscience data will remain primarily as it is. I don’t think this is the case. We will see radical changes in human data consumption as machines become able to do more with the data without human intervention. The suggestion to use “standardized table formats” in relational databases is good but what I think would even be better is to focus efforts on getting the data deployed to the Semantic Web. Nevertheless, these two goals are not mutually exclusive.


Other related blog posts:

Mapping the Brain’s Connections: the Connectome

NIF: When You’re Looking for Neuroscience Resources Including Data

NIF: Neurons, Models, and Grants

NIF: Better Literature Search