Notes from the Biomass

The W3C announced a launch of a Semantic Web Health Care and Life Sciences Interest Group today.

The Semantic Web Health Care and Life Sciences Interest Group is designed to improve collaboration, research and development, and innovation adoption in the health care and life science industries. Aiding decision-making in clinical research, Semantic Web technologies will bridge many forms of biological and medical information across institutions.


What looks like straight from a buzzword generator might substantially enhance the current (babylonic) state of biological databases. Greg over at Nodalpoint recently summarized how RDF, SPARQL et al could help for the integration of bioinformatics resources, including practical problems and support through larger communities (or lack thereof).

There is a substantial increase in such techniques recently and I would not be surprised if they will finally deliver what bioinformaticians have been waiting for (and trying to achieve with other approaches) - integrating the vast amounts of biological data in a flexible way.

One of the challenges to such a system that the W3C cannot address is the availability of the information in a stable and usable form - the EBI can allocate the resources to offer Uniprot in XML but a small lab performing a high throughput screen usually does not have the necessary skills in the lab and will not put their data online in a highly abstracted way unless it hinders publication. On the other hand, I don't expect that journals will raise their standard for publications soon - after all, the format of the information is a lesser part of value of a research publication and the abundance of data is only comparable to those of the possible standards.

The home page of the BIND database now carries a grave statement from its PI, Chris Hogue, which explains that their last dollar was spent on November 16th and that the last thing the team will do is to maintain the status quo of the web servers. BIND is a database collecting protein-protein interactions, hosted by the Mount Sinai Hospital, Toronto, Canada.
Many smaller databases face the same problem - the Postdoc or grad student on the project leaves and funding is hardly ever available for the development of a database, let alone for maintenance. Competition from larger institutions such as NCBI or the EBI often spoil the efforts of many years of development.

However, BIND is not a small databasea and lists more than 100 programmers and curators and received CDN 29$ in public funding in 2003. I have to admit that I never quite understood why one would need such a major effort for a database of protein-protein interactions and can imagine why a project of this size is challenged by other scientists.

BIND is not the first data major database to run out of funding. The situation was similar for the GDB which lost funding in 1998. It was transferred from Johns Hopkins University, Baltimore to the The Hospital For Sick Children, Toronto and subsequently to RTI. The most important database running out of funding was probably Swissprot in 1999(?), which both managed to commercialize its data (thanks to the New Economy biotechs which floated at that time) and attracting support from the EBI.
Chris Hogue selected a few editorials papers covering the situation in the media and promises to continue to report on the situation in his blog.

[Via public rambling]