May 12, 2017 | In August 2015 when Deep Genomics first launched, co-founder Brendan Frey made a prediction. “There’s a sea change coming,” he told Bio-IT World. “People are going to be focusing now on the machine learning component and trying to understand what the genome means, not just sequence a bunch of genomes.” It’s been nearly two years, and he’s been proven right as more and more efforts arise to apply deep learning, machine learning, and artificial intelligence to medicine.
But maybe Frey isn’t wholly prescient. Earlier this month Deep Genomics announced a shift in company focus. Genetic testing was the Deep Genomics business plan in August of 2015; today the company is working to develop genetic medicines.
The company’s technology and foundation haven’t changed. Frey’s vision is to use computer science to accurately model what’s going on in cells and how disease arises from mutations. “Closing the genotype-phenotype gap means understanding how mutations impact what’s going on in cells and how that impacts diseases, whether that’s cancer or Alzheimer’s Disease,” Frey told Bio-IT World earlier this week. Detecting mutations is the first step; figuring out what to do about the mutations is the second part.
But genetic testing as an industry is bound in regulatory constraints. “What we found is that the genetic testing community is very conservative and a lot of issues need to be sorted out. They’re political issues; they’re insurance issues; they’re FDA issues. Sorting that out is not something that we want to focus on in the short term. As the community moves forward and those issues get resolved, then we’ll re-engage with the genetic testing community. But right now we’re focused on genetic medicine,” Frey said.
Deep Genomics intends to understand disease starting with its genetics, and then rationally develop drugs to target the genetic underpinnings of disease. Frey believes the regulatory landscape for that type of genetic medicine is much more fertile, citing the 21st Century Cures Act.
“If you look at pharmaceuticals and therapeutics… everybody realizes that this $2.8b per drug [cost] to produce a drug doesn’t work and an 85% failure rate is just not acceptable. Many people are suffering because of this. So the regulatory constraints are dropping; it’s becoming easier and easier to develop drugs.”
Into The Wet Lab
You might expect that Deep Genomics would simply offer its deep learning platform to companies and groups as software-as-a-service or platform-as-a-service. Frey laughs when I ask him about it; he understands the assumption. But from the beginning, Deep Genomics has employed both computer scientists and cell biologists. The Deep Genomics platform has already identified genetic medicine candidates and the company is pursuing options for central nervous system, eye, and liver disorders, validating them now in tissue culture.
Deep Genomics has lab space at JLABS @ Toronto, a 40,000-square-foot life sciences incubator sponsored by Johnson & Johnson Innovation that just celebrated its first year. Frey is also making good use of the science-on-demand capabilities now available. “Nowadays, over the internet you can order compounds… companies will synthesize the compounds and do the chemistry for you. Companies like Transcriptic enable cloud labs, that allow you to do experiments by uploading basically a computer script.”
The company has previously been funded by angel investors and had revenue from clients and partners. But now Frey is securing additional funding for a Series A round to “massively scale up” the company’s experimental unit. He hopes to announce his own compounds in the next 18 months.
In addition, Deep Genomics is seeking pharma partners. “The path for us is to focus on the early-stage development right now, and collaborate with other pharmaceutical companies and help them get their products out as fast as they can, reduce risk for other pharmaceutical companies,” he said.
Deep Genomics’ vision, Frey stressed, hasn’t changed. “The core idea of Deep Genomics is that the pharmaceutical company of the future is going to look like a computer science company with an amazing team of biologists and chemists and experts in clinical trials rather than a traditional pharmaceutical company with biologists and chemists who are using computational tools. It’s a question of culture; it’ll be a culture of computer science.”
The platform has matured over the past two years. The company has diversified the types of molecular phenotypes it looks at, considering transcription initiation, polyadenylation, and mRNA stability, in addition to splicing errors, and has added protein-related molecular phenotypes.
For the past few months, Frey said, the company has been focusing on how to introduce genetic modifications or therapies to fix various mutations.
“Say there’s a mutation that causes a problem with splicing or a transcriptionally-related problem,” he proposes. “Now what kind of a genetic modification or genetic medication would be needed to fix that problem?”