Humans Out-Compute Supers at Protein Folding
A paper published in Nature this week shows that human creativity trumps computer software, at least in the protein folding arena. A protein folding game, called Foldit, that allows people to puzzle together protein structures, is proving to be quite successful. The game was conceived by scientists at the University of Washington after they got the idea that mere mortals could use their intuition to tweak protein structures in novel ways.
Protein folding has been the domain of supercomputing for some time, given that it’s basically an exercise in molecular dynamics. But the FLOPS needed to simulate proteins of any size is considerable, so even petascale machines have to take some computational short-cuts to predict the molecular structures. Humans, though, can use creativity as a short-cut when problem solving. From the University of Washington announcement:
It turns out that people can, indeed, compete with supercomputers in this arena. Analysis shows that players bested the computers on problems that required radical moves, risks and long-term vision – the kinds of qualities that computers do not possess.
Ars Technica does a deeper dive into why the protein folding software often comes up short when it starts crunching on really big structures:
It sounds simple, but with anything more than a short chain of amino acids, there are a tremendous number of potential configurations to be sampled in 3D space, which can bring powerful computers to their knees. The Rosetta algorithm handles the huge energy landscape it needs to scan by taking big leaps between different configurations, then attempting to minimize the energy by making smaller tweaks. This lets it sample large portions of the structural landscape, but sometimes leaves it stuck: the path between its current location and an energy minimum may take it through a high energy state, which would keep Rosetta from finding the solution.
But it may be only a matter of time before the supers regain the upper hand. The University of Washington researchers are trying to analyze the approaches used by more successful Foldit players, with the idea of trying to replicate those strategies in software.
Full story at Ars Technica