Well it's the last day of the conference and on a whole it's been rather interesting but not overly informative for me personally. Many of the key notes have reiterated may problems in Bioinformatics that I have been aware of for some time now. The key issue that has come up in many talks has been "How do we share data from different databases with different ideas on how to do things wrt. data formats, storage etc." Lincoln Stein's Key Note addressed this problem, comparing the state of bioinformatics now to the states of Italy during the middle ages. The slides for the talk can be found on the O'Reilly Conference web site here. They are in power point format so I can't link directly to slides but I recomend looking at Lincoln's code of conduct for bioinformatics data providers, so very true.
The answer proposed in many talks was to use distributed web services. More on the conference talks can be found at the O'Reilly web site, written by those with better journalistic talents than me.
I'll write up most of my impressions on the conference when I get bact to sunny warm Australia, It's been snowing here, gahh!!
Comments
the open source light
Lincoln and Ewan's talk summaries both make good reading (though as you say, preaching to the converted so far as we are concerned!)
There seems to me to be a contradiction in our attitudes to data. On the one hand, we complain about the diversity of data formats and proclaim the need for a universal standard. On the other, the whole open source philosophy is that there are many ways to do things and we have the choice of the one that suits us.
I'm beginning to think that we don't need a universal standard (how would we ever agree on how to implement it anyway). Perhaps part of being a good bioinformaticist is the ability to deal with diverse data formats easily (and this is where Lincoln was referring to open source libraries like Bio*, I guess). In the end, it shouldn't matter if it's GenBank, Fasta, ASN.1, PDB, XYZ or whatever, it's your skill in transforming it to something useful and analysing it which counts.
Possibly the only solution
Getting biologists/bioinformaticists to decide on a universal data format would be on the one hand amusing and on the other potentially dangerous (physically, we tend to be passionate about these things!). So open formats and open source libraries to deal with the different data formats seems to me to be the only solution.
Also if you find youself creating a new biological data format stop and think "Is there already one out there that suites my needs". I have been looking into how to describe protein-protien interaction networks for my own work. I decided that using an existing markup format (SBML in this case) would be more preferable for a number of reasons. The first being that tools exist to deal with the chosen format (specific xml tools), also greater community support and compatiblity with other software that uses the same format.