SFU researcher receives Michael Smith Health Research Award for work harnessing machine learning to advance genetic sequencing

SFU Statistics and Actuarial Science researcher Lloyd Elliott has received a Michael Smith Health Research BC Scholar Award to support his work using machine learning to understand how genetics affect brain function and neurodegenerative diseases.

“There hasn't been a lot of success yet in translating brain imaging genetics into improved clinical outcomes, but this is going to change” Elliott says. A key challenge in this field has been finding methods of understanding the incredible complexity of the brain at a minute level.

The main tool for measuring the brain in humans is magnetic resonance imaging (MRI), which can provide images of the brain using three dimensional “pixels” known as voxels. “This is useful for giving us a big picture about what's happening in the brain, its structures and connections, but it hasn't led to any deep understanding of genetic neurodegenerative diseases,” Elliott says.

His work applies machine learning to better understand voxels, which can contain around a million neurons. Understanding the brain at this level of detail will help researchers learn how genetics affect various aspects of the brain, as well as genetic links to neurodegenerative diseases and addiction.

Elliott notes that deep learning has already led to breakthroughs in the fields of molecular biology and genetics. However, it has been slower to produce results in neuroscience because the number of participants in studies so far has been small, and the field has relied on classical statistical techniques which break down when analyzing voxels. A larger dataset of genetic samples is needed to make machine learning effective, and new ways of interpreting that data.

“The field is moving fast,” Elliott says. “Consortia such as UK Biobank are working to brain image one hundred thousand participants. Deep learning is going to be more successful in brain imaging in the coming years, and may start to help with diagnosis.” 

“As the puzzle gets put together, we have a chance at identifying potential drug targets,” he says. “I'm confident that within the next 10 years, a drug or therapy will be developed based on these puzzle pieces.”

In addition to his work in neuroscience, done in collaboration with the Wellcome Centre for Integrative Neuroimaging at the University of Oxford, Elliott is a member of the Genetic Epidemiology Sub-Committee of HostSeq, a genetic sequencing project within Genome Canada’s Canadian COVID-19 Genomics Network (CanCOGeN). The project aims to sequence DNA samples from 10,000 Canadians infected with COVID-19.

In a team led by Dr. Steven Jones at the BC Genome Sciences Centre, Elliott is working with SFU bioinformatician Elika Garg to process HostSeq’s genetic data to better understand how the virus affects different people.

“One reason why COVID-19 is so difficult is that the severity is so variable from one patient to the next,” he says. “We're looking for genetic mutations that modulate severity, and we're trying to untangle the connections between severity and comorbidities such as diabetes.”