Social machines at ISMB 2012

This year's 20th annual conference on Intelligent Systems for Molecular Biology (ISMB) was a busy one for me and the rest of the Su Lab.  As a group of 5, we were responsible for 4 oral presentations, three posters, and the administration of one special session.  Keeping that all together while catching up with old and new friends was a fun, though exhausting experience.  And nevermind trying to follow the ISMB twitter stream!

Very briefly, we presented as follows:

• Chunlei Wu on BioGPS - slides, poster
• Erik Clarke on a Task-based evaluation of the Gene Ontology and the human annotations (which one best paper for the Bio-Ontologies session!)
• Salvatore Loguercio on Games for gene annotation - poster, presentation, games
• Myself on a new game for building better class predictors - poster, game
• Andrew Su on Crowdsourcing human gene annotation with the Gene Wiki.

You may notice a fairly consistent theme here (with the exception of Erik's award-winning contribution).  The idea of 'community intelligence' was also explored in the four invited presentations in the special session that I helped to organize.  (Thanks to Alex Garcia for the original idea and rabble rousing that lead to this session's conception nearly two years ago).  At that session we heard from Alex Bateman on the integration of Wikipedia with RFAM, PFAM, and traditional publishing, Alex Pico about lessons learned in the WikiPathways project, Andrew Su about the Gene Wiki initiative, and finally Firas Khatib on the protein folding game Foldit.

Of all the very impressive and interesting presentations, Alex Pico's stood out for me.  To keep this short, I'll leave the specific recap on the other projects to your Googling and finish this with a couple thoughts that percolated from Dr. Pico's perceptive presentation.

A Pico Lesson

WikiPathways is doing great right now.  They have a very rapidly growing user base and are on their way to becoming the de facto standard resource for pathway information.  Of particular interest is that more than 20% of its registered users have made edits to pathways.  That is an astoundingly high rate of user-to-editor conversion.  We claim great success with the Gene Wiki and I am fairly certain that our ratio is less than 1% (though its difficult to tell exactly as the definition of 'user' is fuzzier).

So, how are they making it work?  One of the things that Alex emphasized in his presentation was that WikiPathways was created to solve their own problems in collaboratively editing and sharing pathways (as part of the GenMapp project).  It would have been useful and used (by them) even if no one outside of their research group ever got involved.  The fact that it has been taken up by a broader community is a very valuable, but secondary effect.  This basic idea was echoed in the twitter echoes of Carole Goble's talk (I missed hearing it directly) and resonates yet again with the inescapable Del.icio.us lesson.  Personal value precedes network value.  

As we forge ahead into the realm of Community Intelligence, we need to keep that lesson foremost in our minds.  When we are thinking about games, that means that the game actually has to be fun, really fun!  Time will tell if we can cross that threshold...