After several years in the private sector exploring the unknown origins of neurodegenerative brain disorders such as Alzheimer’s, Chris Ellis thinks one of the keys to solving the mystery is at Oak Ridge National Laboratory: the world’s most powerful supercomputer.
Ellis, a computational microbiologist in the Biosciences Division, is hoping to prove a theory that’s been getting more attention in the biomedical field—that microbes may play a key role in the development of Alzheimer’s and other neurodegenerative diseases.
Ellis and colleagues at ORNL and The University of Tennessee Health Science Center have formed the MIND Consortium—the Microbiome of Idiopathic Neurodegenerative Disease—to determine how the body’s microbiome may be associated with disease states like Alzheimer’s, Parkinson’s disease and ALS.
The consortium’s first project is to collect samples of blood, saliva and stool from 60 patients—15 with Alzheimer’s, 15 with Parkinson’s, 15 with ALS and 15 control patients. The samples will be painstakingly analyzed at ORNL for the identification and metagenomic/metatranscriptomic profiling of microbes.
The data, which Ellis expects to reach as much as 45 to 50 terabytes, will be fed into the Summit supercomputer—part of the Oak Ridge Leadership Computing Facility, a DOE Office of Science User Facility at ORNL. The team will deploy machine learning and explainable AI with the goal of finding how the microbiome influences the onset and progression of neurodegenerative disease.
The researchers hope to discover whether microbes may be colonizing the brain. Their working hypothesis is that a microbe or microbes are the causative agent of neurodegenerative disease, in turn triggering the formation of beta-amyloid plaques implicated in the disruption of neuron signaling in Alzheimer’s patients.
“There have been a lot of recent publications that are giving tantalizing clues that the microbiome is playing a significant role in these disease states,” Ellis said. One paper suggests that P. gingivalis, the bacteria linked to gum disease, can travel from the mouth to the brain and destroy nerve cells. Others implicate the herpes virus, which researchers suggest may be triggering plaque formation in the brain as a defense mechanism.
For Ellis, the work is personal. As with many families, his has been touched by the slow decline of an elder with Alzheimer’s. Ellis lost his grandmother to the disease—the same grandmother who gifted him with his first computer as a boy, and who made him promise to use it for something besides video games. Ellis kept his promise by learning to program.
That same year, in fifth grade, Ellis also received a microscope for Christmas. He started using the 16-kilobyte Texas Instruments computer to record the simple observations he made with the kid-sized scope. “It’s not too surprising that I grew up to be a computational microbiologist,” Ellis laughed.
He earned both his BS and PhD in microbiology at North Carolina State University, with a minor in biotechnology, and received a National Institute of Health Biotechnology Fellowship. As a postdoc at Duke University, Ellis studied the microbiome of extremely premature infants. He then worked for five years at Eli Lilly and Company leading the microbiology group in the firm’s Bioproduct Research and Development organization.
Deep learning and the soil microbiome
Ellis is working on two other major projects involving the microbiome—these within the bioenergy sphere at ORNL. As part of the Plant-Microbe Interfaces Science Focus Area, he is applying his knowledge of computational biology and microbiology to the microbiome of the rhizosphere and comparing it to nearby soil around the poplar tree, identified as a key crop for the development of advanced biofuels.
He and his colleagues, working as part of the Center for Bioenergy Innovation at ORNL, have also received a 10-terabyte sequencing grant from the DOE Joint Genome Institute to analyze samples of the rhizosphere and soil of switchgrass—another target bioenergy crop. It is expected to be the deepest soil microbiome work to date. Ellis explained.
“The microbial diversity of soil is remarkable. We have only cultured a small fraction of the microorganisms that live in that environment and we understand even less about the functional role of the genes they encode. This deep sequencing effort made possible by JGI and CBI will give us greater insight into both the structure and function of the microbiome in soil broadly and switchgrass specifically,” he added.
“I'm really interested in marrying multi-omic processes such as metagenomic, metatranscriptomics, metabolomics, and proteomics together and looking for patterns in the noise using machine learning or explainable AI, at a scale that no one has tackled before,” Ellis said.
It was the supercomputer capabilities that first attracted Ellis to Oak Ridge.
“Near the end of my time at Lilly, I came to realize that if we really wanted to pursue this kind of AI work on the microbiome and do something very ambitious, we were going to need a very big computer,” he said. Ellis has been at ORNL for about two years, during which he has connected with colleagues from multiple related disciplines to propose and fund the AI projects based on microbiome sampling.
“I work with terrific people who are great collaborators here,” Ellis said. “With the multiple disciplines in the microbial area represented at ORNL, it’s like a microbiome itself.”
“The potential here is tremendous,” Ellis emphasized. “And that’s because the lab had the foresight to make this incredible investment in computational infrastructure. As science moves forward, that is really becoming the bottleneck for many different industries. But here at ORNL we can dream big, because we have the resources to be able to dream big.”
With its premier capabilities and expertise, ORNL is “uniquely positioned to make a big impact in these very important disease states,” Ellis said, referring back to the nascent MIND project.
“My grandmother fought the good fight, but we lost her little by little. So many people I’ve met over the years have the same kind of experiences and have told me they’re really hoping for a breakthrough,” Ellis said. “Everyone knows for the patient what a debilitating disease this is, but few people understand how debilitating it is for the family and the caretakers. And so I'm hopeful that we'll find an association that can lead to a breakthrough.”