Startups

Using deep learning and artificial intelligence to map the genome and predict disease

Deep Genomics, a new University of Toronto spinout, uses artificial intelligence and deep machine learning to try to mimic the way in which the cell works - and predict whether a person will get a disease or not.

 

We have another strong contender in the pursuit of a “Google of genome mapping.”

Deep Genomics, a fresh new University of Toronto spinout, is combining deep machine learning techniques with artificial intelligence to study the human genome. It’s building out a database in which a user can type in a combination of mutations found in a patient – and it’ll spin out the likelihood and severity of a patient getting a disease.

“We use machine learning to try to mimic the way in which the cell works – and predict whether a person will get a disease or not,” CEO Brendan Frey said in a phone interview. And he describes it as “something akin to a Google search engine for genomics.”

This could make Deep Genomics a tantalizing new player in precision medicine.

We already have a searchable database of mutations, but what makes Deep Genomics’ approach unique is that it’s opening up a genome-wide database of more than 300 million potentially disease-causing variants, “most of which are in regions of the genome that can’t be examined using other methods,” Frey said.

In essence, Deep Genomics builds out deep learning algorithms that study the genome – figuring out how likely a mutation in an individual is likely to cause a problem.

“Our algorithms are much closer to what’s going on in actual biology than anything else out there,” Frey said.

Frey brought up the BRCA1 and BRCA2 tests: In some women, carrying these mutations will lead to breast cancer. In others, however, the gene does nothing. At present, we can test if a woman has that gene – but we don’t have a good idea whether it’ll manifest itself. Deep Genomics’ algorithms are meant to fit these puzzle pieces together.

Machine learning has been wildly successful in companies like Facebook and Google, and in applications like computer vision, text processing and speech recognition. Part of the reason it’s been easier to unravel these processes with a computer is because we, as humans, already do them ourselves. We read text. We understand speech.

“Understanding the text of the genome is particularly challenging, because we don’t know how to do it ourselves,” Frey said. “We’re asking the computer to do something we don’t know how to do.”

That’s where the artificial intelligence comes in. Frey’s work starts from the ground up – teaching the computer how to read the basic code, associate them with the corresponding RNA and protein outputs, and then building out from there into identifying the manifestation of a disease. There’s a decade-plus of work in Deep Genomics’ platform.

The company has “significant angel funding,” though it’s not disclosing the amount. It’s also got “easily over a dozen different VCs, billionaire entrepreneurs, philanthropists and incubators line up and interested” for the next round, though again, Frey isn’t saying who.

The startup already generates revenue from two kinds of clients – R&D companies working in the longevity space like Human Longevity Inc. and Calico, and clinical labs that are licensing work from Deep Genomics on a per-sample basis.

 

Shares0
Shares0