Mar 3, 2017

Researchers Debate the Future of Genomics

Education, Faculty & Staff, Research
Professor Stephen Scherer at Genomic Medicine Canada 2017
By

Jim Oldfield

Professor Stephen Scherer at Genomic Medicine Canada 2017

“How does a biologist define big data?” That was one of David Glazer’s questions for the audience at the Genomic Medicine Canada symposium this week, hosted at The Hospital for Sick Children by the University of Toronto’s McLaughlin Centre.

The answer? “It doesn’t fit into Excel.” Glazer dropped the line as a joke, but his point was serious: in the last few years, many biologists have turned to new software and infrastructure that can handle the vast streams of data generated by genomics.

As an engineering director at Verily (formerly Google Life Sciences), Glazer is developing new ways to capture and share genomic data. But he came to Toronto with another message in his keynote address: machine learning can produce great insights from data we already have.

Machine learning is a type of data analysis in which a computer develops the ability to recognize patterns and perform calculations based on repeated inputs. The greater the number of inputs, the more accurate the analysis and outputs. Google’s self-driving cars and “smart reply” feature in Gmail are two examples, but Verily is now applying the concepts behind these and other popular applications to the life sciences, with remarkable results.

Last year Google showed that machine learning could teach computers to diagnose diabetic retinopathy as well as and in some cases better than ophthalmologists. And, said Glazer, the ability of a machine to learn from repeated exposure to microscopic lab images and then predict the appearance of cellular markers or signals of cell death — without staining any slides for contrast — may be just around the corner. “It could open up a whole set of quick, non-invasive analytical possibilities,” he said.

The symposium offered Glazer and scientists from across Canada and other countries to share their findings in genomics and big data, and to discuss how those technologies are starting to bring about precision medicine for patients.

Professor Mark Caulfield from Genomics EnglandProfessor Mark Caulfield spoke about Genomics England’s effort to sequence the whole genomes of 100,000 patients and their families in the United Kingdom. He stressed that genomics should be done without the encumbrance of national borders, so that researchers around the world can share and build on the data and treatments emerging from whole genome studies.

Professor Stephen Scherer, who organized the event, spoke about his research on autism spectrum disorders and a Toronto-based, open-science project with Verily and other partners called MSSNG, which aims to sequence the genomes of 10,000 autism patients. Whole-genome sequencing, said Scherer, should become the standard-of care-diagnostic test in autism.

The MSSNG project has already yielded remarkable insights into the individual genetic makeup of each patient’s condition, re-enforcing the idea that autism is really a collection of conditions based on many genetic variants. “We’ve come a long way, and our data is starting to inform on medical management and care pathways,” said Scherer, who is the director of the McLaughlin Centre and a professor of Molecular Genetics at U of T as well as a senior scientist at SickKids. “It’s also refining familial recurrence risk and enabling us to start identifying more appropriate pharmacologic interventions for certain families.”

The day ended with a panel discussion and a plea from former senator James Cowan, who encouraged attendees to contact their local MPs about recent Liberal party amendments to Bill S-201, genetic non-discrimination legislation set for a final vote in the House of Commons next week.

The event was sponsored by the McLaughlin Centre, and Canada’s Genomics Enterprise and partners.