Inferring Single-Cell Regulatory Networks in Neurons Derived from Veterans Afflicted with GWI
Abstract
Overarching Challenges: We are at the intersection of two revolutions brought about by technical breakthroughs. One is the single-cell technology, and the other is machine learning. The proposed study is established on these two pillars of technical revolution, along with stem cell technology regenerating human neurons. We will apply all these jointly to the GWI research, addressing three FY20 GWIRP Overarching Challenges: (1) Diagnosis: Better define and diagnose GWI, (2) Determinants: Validate exposures associated with GWI and impacts on organs and systems, and (3) Consequences: Determine the association of GWI with greater risk for developing neurological diseases. Objective and Rationale: Neurons are specialized cells transmitting information within the nervous system. The cumulative activity of neurons, with their myriad interactions, differentiation histories, and environmental exposures, gives rise to a condition, which is specific to each individual. It is an enduring goal to catalog states of neurons, to understand how neurons activate, how they vary between individuals, and how they fail in disease such as GWI. In the last few years, single-cell technologies have profoundly changed the way research is done in biomedical sciences. Especially, the latest single-cell RNA sequencing (scRNAseq) allows us to measure the transcriptional activity of thousands of genes per cell for thousands of cells at a time, providing a quantitative and ultrahigh-resolution snapshot of cell and molecular states composing a human tissue. In the proposed study, we will use scRNAseq to map all expressed genes’ activity levels across thousands of neurons that are derived from Veteran donors who are either afflicted with GWI or not. We will compare these single-cell gene expression maps between samples to figure out where the most significant differences are and what genes are involved. The comparison is not straightforward though—the neuronal system is driven by intricate interactions among the complex array of molecules, including neurotransmitters and gene products, that comprise the neuronal cells. Our solution is to construct single-cell gene regulatory networks (scGRNs) before comparing them. Characterizing scGRNs constructing using data from many neurons will provide crucial information for understanding how these basic working units interact with each other and work together. The proposed study focuses on applying scRNAseq to neurons derived from induced pluripotent stem cells of Veterans with and without GWI. We will use scRNAseq to obtain transcriptome information from thousands of neurons of each sample. With this unprecedented rich information, we will apply advanced machine learning (ML) to identify specific neuronal gene expression activities implicated in GWI. We will reveal abnormal neuronal circuit function associated with GWI. Our ML method has been specifically designed for comparative network analysis for GRNs from scRNAseq data. The use of these advanced ML algorithms in scRNAseq data analysis is essential. This is because scRNAseq suffers from technical limitations such as the insufficient amount of message RNAs in cells resulting in missing data. The ML algorithms we used are adapted from other fields. For example, tensor decomposition (TD) is a technique used in denoising video data. We use TD to denoise matrix presentations of GRNs. Manifold alignment (MA) is a technique used in automatic machine translation. We use MA to align GRNs to find similarities and differences between Veterans with and without GWI. The whole analytical workflow we developed is highly innovative and powerful. In preliminary experiments, we applied our ML method to real scRNAseq datasets from mouse neurons. We showed that our ML method could identify highly specific gene expression programs in neurons involved in the aging process as well as the acute response to morphine stimulus. The rationale underlying the proposed study is that the l
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Dec 05, 2021
- Source ID
- W81XWH2110171
Entities
People
- James J. Cai
Organizations
- Texas A&M University
- United States Army