Sponsored Post

“Life sciences are the biggest of big data”: a conversation with Cray’s Ted Slater

Meteorologists usually enlist the help of a huge supercomputer when they run their weather models.

Meteorologists usually enlist the help of a huge supercomputer when they run their weather models. Chances are that supercomputer is one made by Cray, a global supercomputing company based in Seattle, Washington.

“So when you watch ‘The Today Show’ and you see the machine that predicts weather for a continent, that’s usually a Cray doing that. When you check the weather on your phone, there’s a really good chance the information came from a Cray,” said Ted Slater, Cray’s Global Head of Healthcare and Life Sciences.

Increasingly, more advanced computing tools are making their way into the healthcare sector, as hospitals and medical professionals try to bring data to bear to treat patients more efficiently, develop drugs more quickly, and generate more effective therapies for diseases like cancer. That’s how healthcare fits into the nuts and bolts of Slater’s job: The bass player who has a background in molecular biology and computer science talks to organizations and individuals doing scientific and medical research with computers, and then figures out if one of Cray’s products, like a supercomputer, might help advance their research. How exactly this type of technology is changing the medical field was the subject of an interview Slater gave during MedCity’s recent CONVERGE conference in Philadelphia.

Before we get to healthcare, let’s back up to the meteorology point, because I think that’ll be relevant. What’s a supercomputer got to do with telling me whether it’s going to rain?

When you’re predicting weather, what you’re working with is a really big mathematical model you need to churn through really fast in order to turn it around in time to make a reasonable weather prediction. That’s what Cray does really well.

And churning through data fast matters to healthcare because?

The reasons are legion. But really, it’s generally recognized that big data is out there to stay and it just keeps getting bigger all the time. Life sciences are the biggest of the big data — healthcare and life sciences sets the standard for the kinds of data that comes in. And in healthcare, where we’re dealing with millions and millions of patients and all kinds of sensors and technologies … you generate enormous amounts of data. And there’s an enormous variety of ways to use those data. For example, at a major pharmaceutical company, you might be very interested in modeling in three dimensions over time what it looks like when a putative drug molecule binds to a target in a biological system. Well that’s all done mathematically, and typically it’s a mathematical model similar to a weather model.

I see. So, technology enables that sort of 3D representation, for instance. But that’s just one way computing is growing in the healthcare field.

The really big thing is precision medicine … It turns out it’s really important to know what a patient’s genome looks like. What’s the particular complement of mutations that they have that defines their disease and defines how they’re going to respond to treatment? That’s obvious from the sequencing itself. Next-generation sequencing, that’s the way you generate those data. And it’s an enormous amount of data, all of which has to be analyzed in certain ways to generate a picture of a patient [for a physician] to recommend appropriate treatment.

And something like a Cray supercomputer can do this because they have machine learning capabilities?

Any computer can do the deep learning stuff. The difference would be that you can run these kinds of things faster on a supercomputer. … It can help researchers explore their data and make scientific decisions rather than waiting their computation to end.

Why is that important?

We’re giving researchers the time they need to really look at their data and do the things with them that they’ve almost been trained to not even think about because their compute environment won’t sustain it. You really cannot do some of the things you want to do because the computer architecture won’t handle it. So in that case a Cray is necessary. … You’re trying to enable a researcher to go through the data they have, see the results of computation, and make a decision about what they want to do next without any limits on what they’re going to try.