Hypothesis Testing for Connectome Comparisons: A Statistical Analysis of Bilateral Symmetry
Thursday, March 3, 2022
5:00 PM-6:30 PM
BIOMED Special Topics: Neuroengineering Seminar Series
Hypothesis Testing for Connectome Comparisons: A Statistical Analysis of Bilateral Symmetry in an Insect Brain Connectome
Benjamin Pedigo, PhD Candidate
NSF Graduate Research Fellow
Department of Biomedical Engineering
Johns Hopkins University
Nanoscale connectomes - maps of neural wiring at the single-neuron level - are increasingly rich sources of data for neuroscientists. Many of the potential uses of connectomes revolve around the ability to compare networks. For example, comparing connectomes could help explain how neural wiring is related to individual differences, genetics, disease, development, or learning. However, the problem of making statistical claims about the significance and nature of differences between networks is an open area of research. Here, we investigate this problem of comparing two connectomes via a case study of the Drosophila larva brain connectome, asking whether we can say that the two hemispheres of this brain network are significantly different.
We describe several ways that one could formalize this notion of “different” as a statistical hypothesis, and we present a test procedure for each of these hypotheses. We find that when viewing the question of bilateral symmetry through a range of simple models, we detect a significant difference between the brain hemispheres. These tests suggest that the difference between the hemispheres is related to both an overall difference in connection density and a cell-type-specific effect.
Finally, we describe extensions of these tests based on our recent work characterizing the structure of this connectome: namely, estimating clusters of neurons based on connectivity and predicting single-neuron matched pairs between brain hemispheres. Taken together, these results provide the first statistical characterization of bilateral symmetry for an entire brain at the single-neuron level, while also giving practical recommendations for future comparisons of connectomes.
Benjamin Pedigo is a PhD candidate and NSF Graduate Research Fellow in the Department of Biomedical Engineering at Johns Hopkins University. Ben works in the NeuroData lab, advised by Dr. Joshua Vogelstein and co-advised by Dr. Carey Priebe.
Ben's research focuses on developing and applying computational techniques for network data to help understand neuron-level connectomes. Currently, he is collaborating with Dr. Marta Zlatic and Dr. Albert Cardona's groups to help make sense of the connectome of the larval Drosophila brain.
Ben also co-leads the development of graspologic, a Python package for statistical analyses of networks, with Microsoft Research. He received his B.S. in Bioengineering (Summa Cum Laude) from the University of Washington and has done two internships at Microsoft Research on network analysis.
Catherine von Reyn