Going Deep Into the Human Genome

On the leading edge of artificial intelligence is deep learning, which could unlock the mystery of how our genes encode human life.

 |  Transcript [PDF]

Some of the most complex science mysteries we’re working on today may not be cracked by a human mind.

Brendan Frey, CIFAR senior fellow and professor of computer engineering at the University of Toronto, is working on a new kind of artificial intelligence (AI) called deep learning.

Old AI systems relied on logic (for example, if-then statements) to make decisions. These all needed to be programmed in. By contrast, deep learning immerses a computer system in data, and lets the computer itself look for patterns.

Modern data sets can be massive, and complex interactions can be difficult to interpret and understand. With enough computing power, deep learning could detect useful patterns that unlock what’s driving complicated systems.

This is a growing field of research in Canada. Frey is a co-founder of the non-profit Vector Institute for Artificial Intelligence, a one-of-a-kind institute that will bring together leading AI researchers, acting as a hub and accelerator for startup companies.

Frey is interested in applying deep learning to look for patterns in life itself, probing genetics with an interdisciplinary group he founded called Deep Genomics.

“My group at Deep Genomics is putting together a system, an AI system, that really is allowing us to peer at your DNA, look at your mutations and figure out what’s wrong and how to treat the disease,” says Frey.

While we now have the technology to rapidly sequence the genome, what comes next remains mysterious; how the genome translates into the expression of biomolecules is not well understood. Frey calls this the genotype-phenotype gap, and closing that gap is needed to understand how genes encode life.

“We’re actually developing new therapies at Deep Genomics,” says Frey, “and that’s what I’m most excited about.”

‹ Previous post
Next post ›

In addition to founding Deep Genomics, Brendan Frey co-founded the Vector Institute for Artificial Intelligence and was a professor in engineering and medicine at the University of Toronto. He has made fundamental contributions to the fields of machine learning and genome biology, both in research and in industry. He led the team that developed a deep learning method for identifying the splicing-related genetic determinants of disease, which was published in the January 9, 2015 edition of Science Magazine. In the past twenty years, he has co-authored over 12 papers in Science, Nature and Cell, including one of the first papers on deep learning (Science, 1995). Frey is a co-inventor of the affinity propagation algorithm and of the factor graph notation for graphical models. He has consulted for over a dozen machine learning-powered companies, has served on the technical advisory board of Microsoft Research, holds seven patents, and has served as an expert witness in patent litigation. Frey’s former team members include entrepreneurs, industrial researchers, and professors at highly recognized centers in Canada, the United States, England and Europe.