WIRED MAGAZINE: 17.05 Gamers Unravel the Secret Life of Protein
By John Bohannon Wired Magazine 04.20.09 http://www.wired.com/medtech/genetics/magazine/17-05/ff_protein?currentPage=all As soon as Zoran Popovic' saw the hair, he knew he was looking at David Baker. It was unmistakable: Baker's face is surrounded by an umbra of curls that organize themselves into unpredictable spirals—not unlike the complex protein molecules he studies. They hadn't met before; Popovic' is an expert in graphics, a computer scientist at the University of Washington in Seattle, and Baker is a biochemistry professor with a laboratory a few blocks away. But David Salesin, another computer scientist and a friend of Baker's, had arranged for the three of them to meet for lunch in a restaurant near campus, because Baker needed help with a tricky problem—and it was exactly the kind of problem Popovic' was good at solving. Baker was the Most Valuable Player in the protein chemistry world's biennial World Series, a competition to see who can predict the shape a protein will fold into, knowing nothing more than the sequence of its constituent parts. It's called the Community-Wide Experiment on the Critical Assessment of Techniques for Protein Structure Prediction, or CASP. With the help of a formidable weapon called rose...@home, Baker's team had dominated CASP since 1998. Like SETI @home, the screensaver that taps spare home computer cycles to sort through radio signals from space, Rosetta farms out computation to volunteered PCs. The 86,000 computers around the world that run it would give Baker the rough equivalent of a 77-teraflop supercomputer for the November 2006 iteration of CASP a few weeks later. But rose...@home was getting stumped by puzzles that Baker thought humans, with their superior spatial reasoning, would solve easily. Baker suggested that users should have some way to tell the screensaver to try a different approach. Popovic' shook his head. "No one's going to care about that," he said. "If you really want people to get engaged, you should put people at the center." Salesin thought so, too—in fact, he thought protein folding would make a terrific computer game. That's why he'd set up this lunch. Baker, thinking about the intense focus he'd seen on the face of his preteen son when playing videogames, agreed. The game they came up with, Foldit, doesn't have orcs or quests or gravity hammers. It simply serves up a multicolored knot of spirals and clumps—a 3-D render of a protein. Players use the cursor to grab, bend, pull, and wiggle the chain of amino acids anywhere along its length, folding the protein into its optimum shape. The only rules are based on physics—opposite charges attract, atomic bonds have limited angles of rotation, and the parts of the molecule that stick to water tend to point outward. The closer your model's properties adhere to those rules, the more points you get. More than 100,000 people have downloaded Foldit since last summer, turning the game into massively multiplayer competition—global online molecular speed origami. And when they came up with potentially accurate CASP protein structures, Baker entered those into the competition. Whoever cracks the hidden secrets of protein folding will push us that much closer to new antibiotics, cancer treatments, and biofuels. Instead of relying solely on computer cycles to accelerate his research, Baker has harnessed neurons and the human urge to play. And if his army of gamers yields a few savants? So much the better, Baker says. "We're looking for prodigies." Hydrogen, the most common molecule in the universe, is just two atoms—each with one proton and one electron. Proteins, the chemical structures that underpin everything from muscles to mothers' milk, are at the other end of the complexity scale: Built on long, kinky backbones of molecules called amino acids, proteins can comprise more than 10,000 atoms apiece. The key to how any protein works is its three-dimensional shape, determined by all the ways its atoms interact. Trying to push two atoms closer when they want to repel is like holding magnets together when they're oriented the wrong way. You can force them, but nature prefers configurations that follow the path of least resistance. In a simple molecule, that path is pretty clear: Water—H2O—is hydrogen-oxygen-hydrogen balanced perfectly in a V-shape at a 104.4-degree angle. This push and pull is inevitable. Physics is destiny. The Challenge The online game Foldit is designed to reveal the shortcuts nature uses to weave a tangle of amino acids, like the one shown here, into a protein. Players click to move pieces around until they fit. http://fold.it/ But the bigger the molecule, the more complex these negotiations become. And proteins are colossal. Of course, there is another way to figure all this out: Shining x-rays through a crystallized lump of protein can help reveal the exact position of its folded-up atoms. But that takes time—just 50,000 protein structures have been cracked since the late 1950s, while the sequences of millions of protein-coding genes have been discovered in the past 10 years alone. To make headway in figuring out what all these proteins do, scientists need a faster approach. So why not just do the math? Calculate the energy of all the different ways a given protein can be folded and find the most efficient. Bang: You're done. But no. Scientists can only estimate the total energy for a molecule as big as a protein, and that's not accurate enough to predict its structure. Even worse, there are more ways to fold a protein than there are atoms in the universe. It's like a combination lock with 1,000 dials. Yet proteins fold themselves into shape in a fraction of a second. No one knows how. Early last year, CASP's organizers hit up laboratories around the world to find proteins whose structures were about to be solved. From these they compiled a list of more than 120 puzzles, which they started posting on the CASP server in May. Popovic' designed an interface for Foldit that renders any protein as a cartoon assemblage of spirals, zigzags, squiggles, and geometric loops. Every part of the protein is movable—push two sheets together and shimmering connectors (representing hydrogen bonds) glue them tight. Try to fit a loop into a hole that's too small and red stars flash at the collision sites. A "wiggle" shakes an entire chunk of the structure to try to get pieces to settle, like dry pasta finding a more compact formation in a jar. Add a chat window and a score tally and you've got yourself a game. The same day Foldit came out, May 8, 2008, an article on the game appeared in The Economist. The ensuing surge of players swamped the server. Working on test proteins for which Baker already knew the structures, folders quickly started to make friends via the in-game chat channel. They shared insights and half-finished puzzles; teams emerged, and collective efforts proved far more successful than any solo folder. A member of the leading team named Jason Kuznicki (game handle: Diderot) set up a wiki that Popovic' adopted as Foldit's official manual. "We even built a mini-Facebook for them," he says. The friendly atmosphere drew about 100 new players a day, and the fierce competition among teams—Freedom Folders, Richard Dawkins Foundation, Folders for Obama—pushed everyone to keep improving. In early June, Baker's team released five CASP proteins to the Foldit community and crossed their fingers. Close to midnight on July 28 of last year, Laurent de Jerphanion (screen name: Dejerpha) stared in disbelief at the multicolored tangle floating on his computer display. The 43-year-old Paris-based marketing manager had been working on puzzle T0461 for several long evenings. There didn't seem to be any improvement left to be made. He was cruising to victory. Then he looked at the scoreboard. He had been overtaken by a 13-year-old American named Cheese. The kid (real name: Aristides Poehlman) had just accomplished an astonishing 20-point jump in a single move—only an hour before the deadline. But de Jerphanion didn't get to be one of the best Foldit players in the world without grit. "À nous deux maintenant," he muttered. Bring it on. On the other side of the world, around 7 pm in Virginia, the Poehlman household was in an uproar. Cheese's parents were folding, too, on computers upstairs. Upon her son's great leap forward, Athena, his mother, typed "Wow! Way to go!" into the Foldit global chat window. More encouragement rolled in from the rest of his team—Another Hour, Another Point—scattered worldwide. But within minutes, de Jerphanion had made more progress on the puzzle and pulled ahead by a point. It was anyone's game. Poehlman's version of the protein looked good. Too good, he thought. No way was he going to make another 20-point jump. That had come from a drastic rebuild of a deeply buried amino acid loop—a risky move. He forced himself to focus on smaller tweaks. He marked two spots on the backbone and clicked an onscreen button to execute a wiggle followed by a sidechain shake. The section of amino acids shivered like a wet dog, but his score didn't budge. Meanwhile, in Paris, de Jerphanion rotated his version of the protein and peered at its innards. One solid improvement would make him unbeatable. He grabbed a loop and nudged it into a gap but pushed too hard. The protein exploded into a Christmas tree of alarms and warning lights, amino acids colliding. He undid the move. Poehlman, too, was trying to squeeze out another point. He spun the protein around and eyed a loop dangling from the end of the largest helix. In pull mode, he guided it along the protein's flank and did another wiggle. The program updated his score. Poehlman banged out a message to his team: "I just pulled ahead by 1 pt." With just a minute to go, Poehlman's parents came downstairs to find their son pacing back and forth in front of his computer, biting his nails with a mouth full of braces. He knew that de Jerphanion could snipe him at any moment. Then the clock ran out. Poehlman danced and hooted while his three-dimensional protein structure uploaded to Baker's server. At 9:40 am an announcement came over the microphone. "Is anyone from the Baker group here?" Three hundred scientists looked around the ballroom. Outside, a cold December wind blustered under a slate-gray sky, but the Hotel Setar Palace, near Cagliari, Sardinia, was warm—and tense. As it did every two years, the CASP community had gathered for the results. The overall scores had just been posted on the Web, and the whole room was beating up the hotel Wi-Fi to get a look. To determine the tally for each puzzle, CASP judges used a formula that compared the guesses against experimentally measured data. In the no-holds-barred series, in which competitors can use human brains, computers, and anything else to solve the puzzles, Baker's team got the highest score in the hardest category, where the puzzles don't look like any known proteins. Everyone pretty much expected that. But the question remained: Had Foldit players contributed any of the winning puzzles? Two of Baker's PhD students, James Thompson and Robert Vernon, groggy from a sleepless night and months of relentless CASP work, finally arrived with the answer. After a brief wrestle with a laptop, they loaded their results. Of the 15 Foldit solutions that Baker submitted to CASP, seven had finished in the money—all of them folded by Poehlman and his teammates. One of their solutions even took first place. A band of gamer nonscientists had beaten the best biochemists. Arguably, though, the real Foldit victory had come a few months earlier. The creators of the game invited the top players to Seattle, seeking their help in making the app better. Popovic' contacted Poehlman's parents. The kid was shocked. "Aristides didn't believe us until we showed him the email," his mother says. "The silent stare he gave us was priceless." Poehlman and his dad, Louis, flew into Seattle late; they played Foldit for hours in their hotel before going to bed—just like at home. At UW's computer lab, Popovic' and his grad students filmed the Poehlmans playing Foldit and interviewed them about their techniques. Louis was exacting in his analysis of how he approached each puzzle, supplying sophisticated justifications for his moves. But when they turned to Cheese and asked him how he knew the way to tweak the proteins—for example, by orienting hydrophobic sidechains toward the protein core—he shrugged and said, "It just looks right." And that is exactly what Baker was looking for. "When I said early on that I hoped Foldit would help me find protein-folding prodigies, it was hopeful speculation," he says. "It's fantastic to see it come true." The next CASP is two years away, and Baker doesn't want to lose Foldit's momentum. He and Popovic' have given the players a challenge: Design a new protein. Baker's lab is developing targets for cancer, AIDS, and Alzheimer's, and the folders' task is to build a small protein drug with the right shape and binding properties. This isn't just an intellectual exercise. Baker says he will synthesize the most promising structures and test them in his lab. These proteins could actually have therapeutic value in the real world, outside the game. And if they do, the Foldit players will share the credit. It might be the first time that a computer game's high score is a Nobel Prize. -------------- John Bohannon (go...@aaas.org) is a correspondent for Science based in Vienna, Austria. -- ================================ George Antunes, Political Science Dept University of Houston; Houston, TX 77204 Voice: 713-743-3923 Fax: 713-743-3927 Mail: antunes at uh dot edu *********************************** * POST TO MEDIANEWS@ETSKYWARN.NET * *********************************** Medianews mailing list Medianews@etskywarn.net http://lists.etskywarn.net/mailman/listinfo/medianews
