Partnership Calls on Supercomputing to Reduce Cancer Drug Time to Market

Massive amounts of health data, supercomputing simulations and partnerships come together to drop the wait times for new viable cancer drugs.

Drugs take years to get through trialing and modeling in the development pipeline, and after that, there are still no guarantees they will make it to consumers.

In fact, one statistic from the University of Michigan Health Lab estimates that 49 out of 50 cancer drug candidates never make it to market.

But a new partnership among pharmaceutical giant GlaxoSmithKline (GSK), the University of California, San Francisco, and two national labs — Frederick National Laboratory for Cancer Research and Lawrence Livermore National Laboratory — known as the Accelerating Therapeutics for Opportunities in Medicine (ATOM) consortium, is looking to change that.

By combining Big Data, artificial intelligence and supercomputing power, ATOM aims to slash the time to market for these lifesaving drugs from six years down to one.

“This is appropriately called a moon shot because we’re trying to do something that could never be done before,” David Heimbrook, the Frederick laboratory’s director, told the Frederick News-Post. “But the overarching need is strikingly clear. It takes six years or more to go from identifying a target in oncology to developing a compound. It’s a very expensive process and it’s a very long process, and in the meantime, patients are dying.”

Heimbrook adds that the cost to develop new drugs is staggering, with estimates of development costs reaching into the billions for a single drug. But, if leveraged correctly, emerging technology can help to shorten this development time.

More Data + More Partners = Less Wait for Cancer Drugs

Through this partnership, researchers from the two labs will have access to the labs’ world-class supercomputers and in vitro biological data for more than 2 million compounds from GSKs enormous database of experimental compounds. GSK will also provide in vitro biological data for more than 2 million compounds that have already failed in development but could help accelerate understanding of what makes a drug successful.

By combining these resources, researchers will be able to test potential cancer pharmaceuticals “in-silico,” or through computer and molecular modeling to determine which might be the most effective.

With the help of supercomputing, computer simulations can analyze “trillions of compounds in a matter of days,” according to the Frederick News-Post, as well as quickly disqualify those that are unlikely to succeed. This can greatly speed up the process and avoid costly drug trials for compounds that may not be suitable.

“It’s completely changing the scale,” Eric Stahlberg, the director of high-performance computing for the Frederick lab, told the Post. “We’re limited in the number of compounds we can evaluate with physical models because we have to have laboratory support for every single one, and often wait for the animals to show signs of the disease.”

“The more data we can get, the better we can predict what will work and what won’t work,” Heimbrook said. “Even if we could get the process from six years down to two years, that would still be transformational.”

Ultimately, ATOM aims to develop a single drug discovery platform that combines several emerging technologies, including AI, supercomputing simulations, data science and more, into a platform that can be distributed to the drug development community.

In the meantime, the consortium is actively seeking new members to fill out its ranks.

The more data we can get, the better we can predict what will work and what won’t work,” Heimbrook said. “Even if we could get the process from six years down to two years, that would still be transformational.”

Yuri_Arcurs/Getty Images
Nov 27 2017