A Complex Network-Based Approach for Detecting and Investigating Colonist Neurons in Drosophila

Authors:
Open Access
Publisher website
Google Scholar
Abstract
The connectome of an organism represents the intricate network of its neuronal connections and is responsible for complex brain activities such as memory, learning, perception and attention. The comprehensive mapping of the connectome of complex organisms is a challenging issue. However, thanks to advances in neuroimaging techniques and computational methods, researchers have been able to reconstruct the complete connectomes of simple organisms. The most complex organism whose complete connectome has been reconstructed is Drosophila melanogaster, commonly known as fruit fly. The complete connectome of the larval and adult stages of this organism has been known for a couple of years. From this knowledge, researchers have proposed a variety of studies to better understand the brain function of this organism. This paper is in this line of research and proposes the identification and subsequent characterization of “colonist neurons”, i.e., a new class of neurons that exert a substantial influence on regions of the brain that are physically distant from those in which they are located. In the paper, we give a precise definition of these neurons and show that they have some extremely interesting properties, including (but not limited to) the ability to be strongly interconnected to form a backbone and the ability to form colonies in distant regions over which they exert their influence. In addition to investigating colonist neurons, we propose a study of their colonies, that is, the set of neurons through which they exert their influence in remote regions of the brain. From a computational perspective, the proposed approach is a graph-based data-mining framework combining spatial network analysis and unsupervised learning to identify unrecognized organizational patterns in large-scale connectomic data. This organization, characterized by local colonies connected by sparse, long-range projections and a backbone of influential neurons, may also provide structural insights for designing and analyzing biologically inspired artificial neural systems.

This publication has 47 references indexed in Scilit: