BIND goes on the offensive?

My knowledge of patent law (U.S patent law, at that) is rather sketchy, but I was surprised to read (via Snowdeal) that Unleashed Informatics - the commercial spinoff from the now defunct Blueprint Initiative - is planning on enforcing a patent it holds covering "a system for electronically managing, finding, and/or visualizing biomolecular interactions comprising a computer system including at least one computer receiving data on biomolecular interactions from a plurality of providers" (full patent here). To be fair, Unleashed has said that it will only target for-profit organizations.

Of course companies have to protect their intellectual property, and I don't have any problem with that. Some points that concerned me, though:

It's a software patent. Fine, patent software ideas so that somebody else can't patent it after you and then take you to court, but enforcing it yourself? Seems dodgy, and it's not like there are hundreds of competitive commercial BIND clones out there (AFAIK). Are UI under pressure from funders, do they just need a quick buck, what?

Surely US-based databases like Reactome fall under the patent - are they effectively at the mercy of Unleased Informatics deciding to be nice to academics? What happens if UI goes bust and the patent rights are acquired by a company not so nice?

Tangled Bank #51

Just a quick post to say that Tangled Bank #51 is up at DBDW. Tangled Bank is a collection of science related blog posts that comes out once a fortnight. While you're there check out Discovering Biology in a Digital World's archives for some more interesting reading.

Standardized testing

There's an interesting, wide ranging review paper in Biology Direct this month that looks at methods for motif discovery. At the end of the introduction the authors point out how difficult it is to perform any kind of quantitative performance measurement on the different algorithms and suggest that there's a need for more standardized testing in the field, citing Tompa et al.'s excellent paper in Nature Biotechnology as a step in the right direction (Tompa compared several modern motif finding algorithms on a carefully selected reference set of transcription factor binding sites).

Broadly speaking I'm a fan of different research groups getting together to establish standardized testing. It can drive innovation - look at all the methods created specifically for CASP or BioCreAtIvE - and there's no doubt that it's handy for biologists coming in from the cold. Imagine that all you want to know is if a particular program that you use has been superceded by newer, fancier algorithms yet: would you rather compare two statistics to the top-ten list on the assessment web page or go off on a tangent chasing references?

Of course, it's sometimes easier said than done. I'm interested in finding candidate disease genes in large regions of interest and there are at least a dozen different algorithms that can help with this, but comparing them is difficult as they're all suited to slightly different circumstances (for example, one works well when there's expression data available but cannot operate otherwise: is this method better or worse than an algorithm that doesn't require expression data but does need, say, GO annotation?). Working out which algorithm will work best on the data that you're interested in becomes obvious if you read all of the relevant papers but it's information that could quite easily be lost in a standardized study.

As a brief aside, check out the reviewer comments from Eugene Koonin on the Biology Direct paper: why don't reviewers keep it short n' sweet like that when commenting on my manuscripts?

On second thoughts, don't answer that.