A Microscopy and Bioinformatics Toolkit for High-Throughput Connectomics Studies

Abstract

It has long been known that the mammalian brain is constructed of thousands of distinct cell types which are individually defined by their gene expression, morphology, connectivity to other neurons, and electrophysical response to stimulus. Understanding how neural networks are structured at a single-cell level, however, have been limited by the scalability of microscopy and data analysis tools used for neuron reconstruction studies and the speed at which connectomes can be created. Recently, the application of Spectral Connectomics has allowed dense by combining brainbow neuron labeling, expansion microscopy and high speed volumetric neuron labeling. In this thesis, I present a broad toolkit which allow Spectral Connectomics to be performed in a higher-throughput manner. First, a high-speed parallel line scanning confocal microscope (plSCM) is described, which allows high speed, multispectral, and volumetric imaging at a rate of 18Gb/s. A custom scalable network-distributed image acquisition framework (SNDiF) is used to allow image acquisition to be performed over a fiber optic network, storing images directly on a computation cluster. This combination is used to collect large speed and high-speed microscopy of living tissue and the mouse brain. A novel scalable image storage format (SISF) was developed to enable quick accession of multi-terabyte images, even when they are stored on archival and supercomputer storage systems. We developed nTracer2, a cloud-based platform to enable PB-scale brain image visualization and neuron tracing in the cloud. nTracer2 utilizes Google Neuroglancer’s web browser-based application for remote client data access. nTracer2 consists of three independent components: a content delivery network (CDN) server that synthesizes the proper views from the raw dataset and streams them to the user end, a database backend that supports parallel data analysis and result curation in the cloud, and a neuron tracing interface that is added to the user’s browser. Finally, I present several software tools for analysis of neuron morphology, including the nGauge Python package. Together, this pipeline allows creation and understanding of biological neural networks at an unprecedented scale, which may have broad implications for our understanding of neurodegenerative diseases. The tools that I create are robustly validated and we commit to share these tools with other researchers in the neuroscience community under the NIH FAIR principles.