Page 17 - MDC Abstract Book & Guide

Abstract
Neurodegenerative diseases such as Alzheimer’s (AZ), Parkinson’s disease (PD), and Amyotrophic Lateral Sclerosis (ALS) affect millions of people, cost hundreds of billions of dollars annually, and yet there is no compelling correlation between genetics and disease. In AZ specifically, the amyloid beta hypothesis has a 100% clinical failure rate despite the investment of tens of billions of dollars and decades of effort by the pharmaceutical industry. Finding the true disease etiology for these neurodegenerative diseases is therefore a compelling and urgent unmet medical need. To address this need we have created the Microbiome of Idiopathic Neurodegenerative Disease (MIND) consortium. We are leveraging world’s fastest supercomputer (Summit) at Oak Ridge National Laboratory to develop and apply new analysis platforms to multi-omic, multi-scale data using data analytics, machine learning, and explainable AI. To accomplish this, we took samples from multiple sites (oral, fecal, blood) from patients diagnosed with AZ, PD, and ALS as well as carefully selected control patients for comparative analysis. Using a comprehensive suite of multi-omic techniques (metabolomic, metagenomic, and metatranscriptomic) we are analyzing a broad range of putative microbial pathogens (virus, fungal, bacterial, archaeal, protists) associated with the onset of neurological diseases and in parallel looking for novel biomarkers for new rapid diagnostics.
Geoffrey Hannigan Merck Exploratory Science Center
Geoffrey Hannigan is an Associate Principal Computational Biologist at the Merck Exploratory Science Center (ESC) in Cambridge, MA. Geoffrey and his colleagues perform research to understand the interactions between the microbiome and the immune system in disease. Their goal is to translate their findings into clinically impactful therapies, biomarkers, and drugs. Before joining Merck, Geoffrey earned his PhD at the University of Pennsylvania while studying the skin microbiome and virome with Elizabeth Grice. He did his postdoctoral fellowship at the University of Michigan with Pat Schloss, focusing on the gut microbiome and virome in Colorectal Cancer.
Presentation Title
Understanding Human Microbiome Functionality, Immune Interactions, and Biomarker Potential Through Deep Learning
Abstract
The human microbiome, the collection of bacteria and other microbes that colonize the human body, has a profound impact on human physiology and disease state. Additionally, the interaction between the host and the microbiome plays a major role in the development and functionality of the immune system. These interactions are driven in large part by the metabolic processes of the bacterial members of the microbiome. One of the mechanisms by which bacteria synthesize secondary metabolites is through biosynthetic gene clusters (BGCs). BGCs are clusters of genes that are physically co-localized in a bacterial genome and together encode the components required to produce a metabolic product. Our group has enhanced the detection and analysis of BGCs through the development of methods including DeepBGC, a BGC detection algorithm based on a long short-term memory (LSTM) neural network. The use of these and other analytical tools in our microbiome studies has allowed us to begin gaining unique insights into how the microbiome may impact immune responses, how those signals might be used in the context of biomarkers, and how that may be leveraged therapeutically to improve treatment efficacy.