March 3, 2017

Fraxinus - a gamer's review

One of the top UK news stories yesterday came from our friend Dan Maclean at the Sainsbury lab over in Norwich - some of you may remember him from his presentation at our second symposium in 2012. Dan and his collaborators have developed a Facebook game to help with research into ash dieback, a disease that is currently placing the UK's substantial population of ash trees under threat. Certain individual trees have been found to be resistant and researchers are now trying to identify the genetic reasons behind that so that they can select the correct, resistant individuals for breeding a replacement population of saplings as the mature trees begin to die off. Understanding the disease itself and producing a treatment would also be a major benefit but at this stage it is protection of the tree stock itself that it is a priority.

Dan's game is called Fraxinus. There was much coverage yesterday of the goals of the project and the innovation of using a computer game to help solve the problem, but what is it like to actually play? I could hardly call myself a seasoned gamer, but last night I thought I'd give it a go.

fraxinus screen shot

Image taken as a screenshot from Fraxinus. Courtesy of John Innes Center.

The basic premise is to align rows of leaves against a pattern at the top of the screen. Align them correctly - by moving them left or right, removing a leaf here or there or inserting a gap or two in the middle, and obtain maximum points. It is hardly a complex concept, but anyone familiar with bioinformatics will realise straight away that what is happening here is the alignment of sequencing reads against a reference genome. Align them correctly and you stand a much better chance of understanding the variations between individuals, and hence the markers which might be associated with disease resistance. Computers do this well but still nothing does it better than human pattern recognition and intuition.

I played about ten rounds - different patterns - before I have to admit I got slightly bored. The concept was fine, but the presentation could do with a little tweaking. For instance, the rows you are asked to align to the pattern are already aligned against each other in most cases, but not against the pattern. It would be better if what was presented was the computer's best guess, or a nearly-guess (modified to push it off by a few positions here and there to prevent leading the player down a particular line of thinking), but the current presentation often puts rows tens of positions off to one side and when you have to move all of them, one position at a time, one click at a time, it becomes tedious quite quickly. Also, some of the rows bear no obvious relation whatsoever to the pattern and it is not clear whether this is accidental or intentional - there is no 'I don't see any alignment here' button, only a 'this is the best alignment I can get'.

Having said that, it passes the time quite nicely! The visual presentation is excellent (save for the minor usability issues above), and its got a pleasant but unobtrusive backing track that makes time go much quicker than it probably would otherwise (there speaks the musician in me). Nobody can fault the aim of the project and it is clear that the alignments produced will greatly help research (assuming that they have somehow been preselected as being the most likely areas of importance - at least I hope they have!), but playability lets it down a little. 

I can be a little harsh sometimes, particularly when I'm reviewing journal submissions, but I can't help but give Fraxinus a 7/10 - not a perfect score, but certainly a very good job.

Topics: alignment, ash, Bioinformatics, Bioinformatics, computer, deletion, dieback, disease, fraxinus, game, gaming, gaps, insertion, intuition, pattern recognition, playability, read, research, sequence, tree, usability