Image via Wikipedia
Here’s an interesting service I discovered during a snoop on the web. PDBalert is a web-based system that alerts users as soon as a pdb structure with homology to a protein of interest becomes available. Users can upload protein sequences of interest and ever Wednesday, when the PDB releases new structures, the service compares sequences to all the new proteins. Simple enough. I suspect many people have their own scripts which do essentially the same.
The thing that jumped out at me was how easy it could be for the PDB to create a service that does the same. You point the service to a source file(s), choose an appropriate algorithm (they could give you some choices), etc. In a perfect world, you could even mash this up with some kind of function prediction engine, etc. The way I see it, more services the better, esp if they can talk to each other. Some day. I still believe there is room in the science space for an API management service which allows developers to build tools upon existing resources like the PDB.
Deepak is absolutely correct. There has to be greater attention to scientific APIs, especially providing users with the abilities to manage these APIs and to perform more complex mashups.
So how about getting a tweet on twitter instead of email, or even have it appear on a Facebook page? Then have it hit pubmed and provide useful papers dealing with the new protein. Why not then directly provide DNA sequence homologies, etc?