More Than Skin Deep: P&G Using GPUs to Study How Skincare Products Get AbsorbedNovember 26, 2013
When talk turned to skincare products at Supercomputering 2013 (SC13) – a gathering of some of the world’s most sophisticated technologists – the audience of largely balding, bearded, middle-aged men leaned in, eager for more.
They were taking in an account of Proctor & Gamble’s effort to model, down to the atomic level, how the body absorbs compounds through the skin’s primary protective barrier. The work can potentially avoid weeks of costly experimental work with live subjects or other substitutes.
This is no small matter for P&G, which owns a range of multibillion-dollar beauty and skin-care brands like Olay, CoverGirl and Max Factor. Its goal in the work is to develop safe, effective products faster and more efficiently.
Speaking during last week’s show at NVIDIA’s GPU Technology Theater, Russell DeVane, a P&G computational chemist, said his team is using Titan, the fastest supercomputer in the U.S., to run simulations down to 50,000-200,000 atoms showing how certain compounds can pass through the skin’s protective barrier, while others can’t – and others cause significant disruption to the barrier.
This involves studying the complex relationship between a compound’s molecules and the stratum corneum – the skin’s primary protective barrier – plus its matrix of lipids which act something like a “mortar” between dead cells. Lipids are the primary path through which compounds can penetrate.
P&G has its own Tesla GPU-based clusters, which provide speed-ups of four to eight times on applications for molecular dynamics work, such as NAMD and LAMMPS. But it turns to Titan at the Oak Ridge Leadership Computing Facility for much larger-scale efforts.
DeVane described efforts to better understand the barriers a compound faces when penetrating through the skin.
“We want to build a model that would allow us to use modeling data, rather than experimental data,” he said. “We can quantify the ability of different compounds to cause disruptions” in the skin’s protective layer.
He noted that even Titan, for all of its GPU-accelerated processing power, can’t simulate the workings of the full skin. But he said his team’s models enable them to simulate thousands of compounds in just a few minutes.
If you missed DeVane’s presentation at SC13, you can view it and all of the other GPU Technology Theater talks here.