My latest Mind and Matter column in the Wall Street Journal is on citizen science:
The more specialized and sophisticated scientific research becomes, the farther it recedes from everyday experience. The clergymen-amateurs who made 19th-century scientific breakthroughs are a distant memory. Or are they? Paradoxically, in an increasing variety of fields, computers are coming to the rescue of the amateur, through crowd-sourced science.
Last month, computer gamers working from home redesigned an enzyme. Last year, a gene-testing company used its customers to find mutations that increase or decrease the risk of Parkinson’s disease. Astronomers are drawing amateurs into searching for galaxies and signs of extraterrestrial intelligence. The modern equivalent of the Victorian scientific vicar is an ordinary person who volunteers his or her time to solving a small piece of a big scientific puzzle.
Crowd-sourced science is not a recent invention. In the U.S., tens of thousands of people record the number and species of birds that they see during the Christmas season, a practice that dates back more than a century. What’s new is having amateurs contribute in highly technical areas.
Protein folding is certainly highly technical. The amino-acid sequence of a protein, determined by a gene, in turn determines the shape of the protein. Though our cells have no trouble following instructions for building the protein, predicting that shape from the amino-acid sequence is fiendishly hard for our minds, even assisted by computers.
The correct result is usually the structure with the least contorted shape, but there are so many ways you can go at each step that you’re soon lost in a 3-D jigsaw of astronomically large possibilities.
This matters to drug designers and organic chemists, because when they want to come up with a protein that will fit some target, they have to predict what sequence of amino acids will give the best shape. David Baker and Zoran Popović of the University of Washington in Seattle had the idea of farming the problem out to people who play computer games. Just like any other game, “Foldit” rewards gamers who collaborate and compete to come up with ingenious solutions to protein-folding problems.
Already Foldit players, who generally have no background in biochemistry, have designed proteins with a good fit to the antigens on the 1918 influenza virus and have helped to figure out the structure of a retroviral protein.
Now they’ve come up with over 100,000 designs for an improved version of an enzyme for catalyzing the Diels-Alder reaction, which is useful in organic chemistry, but for which no naturally occurring enzyme seems to exist. A protein that Dr. Baker had designed didn’t work well. The best of the Foldit players’ designs proved to be more than 18 times more efficient.
A similarly exciting result came last year from the firm 23andme, which markets direct-to-consumer genetic testing and now does research (which it calls “23andwe”) by, in effect, comparing customers with each other. Their first target was Parkinson’s disease, not least because 23andme founder Anne Wojcicki’s husband, Google co-founder Sergey Brin, carries a rare mutation, called LRRK2, that increases his risk of Parkinson’s disease to roughly 50%.
By sequencing many parts of the genomes of 6,000 Parkinson’s patients (and counting) and comparing the data of many of their 120,000 other customers (90% of whom opt in to be part of such studies), 23andme has found two genetic mutations that seem to raise the risk of developing the syndrome. The firm also found that genes at least partly determine when Parkinson’s disease develops.
Recently 23andme said it had found a genetic variant, SGK1, that seems to protect those with Mr. Brin’s mutation against the disease. Assembling and sharing so much data would be a tall order for the conventional pharmaceutical industry. Crowd-sourced science seems to be the future.