Department or Program



It has been proposed that one of the clearest organizing principles for most sensory systems is the existence of parallel subcircuits and processing streams that form orderly and systematic mappings from stimulus space to neurons. Although the spatial heterogeneity of the early olfactory circuitry has long been recognized, we know comparatively little about the circuits that propagate sensory signals downstream. Investigating the potential modularity of the bulb’s intrinsic circuits proves to be a difficult task as termination patterns of converging projections, as with the bulb’s inputs, are not feasibly realized. Thus, if such circuit motifs exist, their detection essentially relies on identifying differential gene expression, or “molecular signatures,” that may demarcate functional subregions. With the arrival of comprehensive (whole genome, cellular resolution) datasets in biology and neuroscience, it is now possible for us to carry out large-scale investigations and make particular use of the densely catalogued, whole genome expression maps of the Allen Brain Atlas to carry out systematic investigations of the molecular topography of the olfactory bulb’s intrinsic circuits. To address the challenges associated with high-throughput and high-dimensional datasets, a deep learning approach will form the backbone of our informatic pipeline. In the proposed work, we test the hypothesis that the bulb’s intrinsic circuits are parceled into distinct, parallel modules that can be defined by genome-wide patterns of expression. In pursuit of this aim, our deep learning framework will facilitate the group-registration of the mitral cell layers of ~ 50,000 in-situ olfactory bulb circuits to test this hypothesis.

Level of Access

Open Access

First Advisor

Castro, Jason

Date of Graduation


Degree Name

Bachelor of Science

Number of Pages


Components of Thesis

1 pdf file

Open Access

Available to all.