CBioC: beyond a prototype for collaborative annotation of molecular interactions from the literature

In molecular biology research, looking for information on a particular entity such as a gene or a protein may lead to thousands of articles, making it impossible for a researcher to individually read these articles and even just their abstracts. Thus, there is a need to curate the literature to get various nuggets of knowledge, such as an interaction between two proteins, and store them in a database. However the body of existing biomedical articles is growing at a very fast rate, making it impossible to curate them manually. An alternative approach of using computers for automatic extraction has problem with accuracy. We propose to leverage the advantages of both techniques, extracting binary relationships between biological entities automatically from the biomedical literature and providing a platform that allows community collaboration in the annotation of the extracted relationships. Thus, the community of researchers that writes and reads the biomedical texts can use the server for searching our database of extracted facts, and as an easy-to-use web platform to annotate facts relevant to them. We presented a preliminary prototype as a proof of concept earlier(1). This paper presents the working implementation available for download at http://www.cbioc.org as a browser-plug in for both Internet Explorer and FireFox. This current version has been available since June of 2006, and has over 160 registered users from around the world. Aside from its use as an annotation tool, data from CBioC has also been used in computational methods with encouraging results.