People who play a simple online video game developed by Montreal researchers are helping crack genetics problems that leave computers puzzled.
"It takes forever if you want to try all solutions and pick the best one," said McGill University computer scientist Jérôme Waldispuhl, explaining why computers have difficulty coming up with a solution. "But humans have this capacity of having intuition [for] where the solution is and get straight to the point."
The solutions that geneticists want to find — and that the game Phylo generates — are similar DNA sequences between two different organisms.
The similarities among sequences are greater and the matching is easier in the lower levels of the game, which involve animals that are closely related, such as humans and chimpanzees. At higher levels, players must perform matches for less closely related animals such as humans and fish. The game is named for the "phylogenetic tree" that describes how organisms are related to one another through evolution.
Players can move to the next level when they achieve a higher score than a computer did on the same problem. At that point, the player's solution is sent back to the researchers.
The sequences used in the game come from the University of California Santa Cruz's genome program, and were chosen because they are poorly aligned sequences within genes that show overall similarities between different species.
Those are starting points for researchers to make discoveries that could be used to treat problems such as cancer or heart, nervous system or immune system diseases.
If many organisms that evolved at different times have the same or very similar genetic sequences, that sequence has probably been very important throughout evolutionary history — it may be the instructions for making an important protein, for example.
"It also means that if you perform a mutation in this region, eventually you can expect that a problem will happen," said Waldispuhl, whose research mainly involves creating computer algorithms to solve problems involving biological data.
Phylo involves shifting rows of coloured blocks, which represent DNA sequences for two different organisms, such as a horse and a dog. The goal is to slide them until they match as closely as possible with as few gaps as possible. The game works because humans have evolved to be good at solving such visual puzzles, said Waldispuhl, who worked on the project with his colleague, Mathieu Blanchette, along with two undergraduate students, Alex Kawrykow and Gary Roumanis.
"The idea was really to make it as fun as possible," said Waldispuhl, who hopes to be able to "crowdsource" the skills of thousands of ordinary people.
7,000 puzzles in 1 day
The game launched on Monday, and as of Wednesday, 1,500 people had signed up to play. They solved 7,000 puzzles on Tuesday alone.
That caused some problems for the servers that hosted the game, which were designed to allow hundreds of connections rather than thousands.
"We've been quite amazed by the number," Waldispuhl said.
The current version of the game was written in Flash for a web interface by Kawrykow and Roumanis. They already had experience creating Flash games, and had good intuition about how to make the game appealing by carefully designing its look and feel, Waldispuhl said.
Waldispuhl contributed the music for the game, a jazzy piano number he composed with a band he was in while pursuing his grad studies in Paris.
Facebook, mobile versions planned
Now the team is hoping to create versions for Facebook and all mobile devices, as well as versions in other languages such as Chinese.
Using video games has already solved some other biology problems.
In August, University of Washington researchers published a paper in Nature showing that players of a video game named Foldit were better than computers at figuring out the best way to fold a protein into a 3D structure based on the laws of physics. The shape of a folded protein is crucial to its function as a lock or key in biological processes.
Now some of the researchers behind Foldit are analyzing some of the most successful Foldit strategies used by humans in order to replicate them in computers.
Waldispuhl said compared with Phylo, Foldit is a fairly complicated game.
"If you're not a specialist or a bit [of a] geek, you won't be able to get into the game," he said.
He believes the simpler problems being solved in Phylo and their wide application in genetic research lend themselves better to a video game designed to generate scientific data.