PgmNr D160: Features and Applications of FlyCircuit Database – From Fluorescent Images to the Drosophila Connectome.

Authors:
C. T. Shih 1,2 ; A. S. Chiang 3,4


Institutes
1) Tunghai University, Taichung, TW; 2) National Center for High-Performance Computing, Hsinchu, TW; 3) National Tsing-Hua University, Hsinchu, TW; 4) Kavli Institute for Brain and Mind, CA.


Keyword: webtools and databases

Abstract:

FlyCircuit (www.flycircuit.tw) is a database containing more than 28,000 image entries of ­individual neurons in the Drosophila brain. The database was constructed for online data archiving, browsing, searching, analysis, and 3D visualization for the single neurons. Using the genetic mosaic analysis with a repressible cell marker (MARCM), the single neurons were labeled with green fluorescent protein (GFP). Also, the birth date during the brain development and putative neurotransmitter for each neuron could be predicted with this transgenic technique. To handle these terabytes of image data, a fully automatic image-processing workflow was introduced to reconstruct the single neurons from the fluorescent raw images to their three-dimensional digital reconstruction.

With the image data in FlyCircuit, the structural connectome of the fruit flies could be assembled at different scales. The mesoscopic connectome was composed of the brain regions called the local processing units (LPUs) and their directed and weighted links. It showed small-world characteristics, hierarchical modular structure, and rich-club organization. The network consisted of five functional modules – olfactory, auditory, left and right visual centers, and the pre-motor center. Major loops with length from three to five were identified. Many loops were found which could sustain the electrophysiological signals for a longer period.

Microscopic connectome composed of the neuron-to-neuron connections was also analyzed. The network also showed hierarchical modular structure. Interestingly, the modules were corresponding to the functional modules in the mesoscopic version, rather than the expected LPUs or neuropils in the brain anatomy. The numbers of loops with length up to six in the brain network were significantly larger than their counterparts in the random networks. Finally, rich club formed from a population of neurons with high global centrality scores was also detected.