phylo.cs.mcgill.ca

Completing a multiple sequence alignment—a way of arranging the sequences of DNA, RNA, or protein to identify regions of similarity—is an age-old problem in genetics. To tackle this task, two bioinformaticians at the McGill Centre of Informatics have employed a somewhat unconventional method. They decided to make an online puzzle game out of the sequencing problem and let players compete with each other on the web to produce better solutions.

That is how the Flash-based game called Phylo (short for phylogenetics) was born. Phylo is built upon the premise that the human brain is inherently light-years ahead of computers when it comes to color matching and visual pattern recognition.

 Jérôme Waldispühl, a bioinformatics expert and one of the project leaders of Phylo, discussed the motivations behind the creation of the game.

“Humans are good at solving, and enjoy solving visual puzzles. With the right setup, interface, and if the science of the game can be abstracted in a way that is fun enough, then anybody will want to play it,” Waldispühl said. “We [could then] have access to a huge number of people who can contribute to solutions that may be better than the existing solution.”

Producing such an alignment is traditionally done using a computationally complex heuristic algorithm, which does not guarantee an optimal solution. There is only so much the computer can quantify before it becomes prohibitively expensive to perform the computation. This is due in part to the sheer size of the genome, which consists of roughly three billion base pairs.

In Phylo, instead of arranging real nucleotides, players align four colourful blocks of squares—representing the four nucleotides of DNA—in a column to find the best alignment between snippets of DNA from two different species. Players compete against the computer, as well as others, to get the best possible score on each puzzle, with scores determined by how closely the coloured shapes are arranged. When a puzzle is solved, the information is analyzed and stored in the University of California, Santa Cruz’s genome browser, a database that catalogs the billions of stretches of genetic information and makes them available for scientists in the form of a usable visual digest. 

Despite the amusing nature of the game, which is complete with a jazz soundtrack, project leaders insist that players solve real sequences and make legitimate contributions to science.  All solutions contributed from players are useful data. 

“When people play casual games like Tetris or Minesweeper, they are actually performing computations,” Waldispühl said. “However, after all that trouble of achieving efficiency and getting the highest score, nothing comes out of it [except] killed time. For us computer scientists, this is a waste.  Phylo provides a good mechanism to ‘recycle’ the computations that are already made by the human brain, from being un-utilized into meaningful data.”

Optimal solutions to sequencing DNA can be pivotal to finding clues about diseases that are have some genetic basis, like breast cancer. According to Mathieu Blanchette, also a McGill bioinformatician and another Phylo project leader, mutations are common when sequencing DNA, but detecting which mutations are potential cursors of diseases requires looking at sequence alignments.

“One powerful way to determine whether a piece of human mutation is relevant is by looking at the corresponding region in the genome of this species across evolution. Geneticists detect whether this type of mutation is well conserved or more variable across evolution in different species.  Good alignment helps in this endeavor,” Blanchette said.

Since Phylo’s inception in November 2010, the results have been very positive to date. Blanchette pointed out that not only are the game’s solutions better than those returned by a computer, but they also give rise to a diverse set of solutions. At the very least, solved puzzles—even non-optimal ones—contribute to better initial data or a starting point that can be fed into a computer for further completion that may lead to better computations.

“We go back to the biologists and ask whether the solutions helps. They have said solutions have been exponentially better,” Blanchette said.

Phylo is taking advantage of the Facebook  and Google+ communities to help spread the word and promote the game. After all, taking on the sequencing of the human genome of three billion base pairs is no small feat.  The project leaders have also completed a version that is available as an app on mobile phones and the iPad. In the future, the team is hoping to build a Phylo open source community at McGill and across the web, where individuals can help contribute to improving the game. 

Waldispühl has always been driven by the idea of attempting to solve scientific problems by involving non-scientific communities.

“My dream is to build a dedicated community that is devoted to building casual games that not only solve biological problems but other suitable problems as well,” he said. “I would like to see more games that are built and played to make a difference in this world.”