Given the diversity of biological information, we strongly believe that a retrieval system can perform better by integrating the links that exist between biological databases covering different areas and different types of data. As biologists identify new genes and gene functions every day, new sequences are stored and new literature is published at an increasing speed. The size of nucleotide sequences databases such as GenBank is growing larger as well as the size of protein sequences, protein structures and biomedical articles databases. The data is often structured and organized according to a project covering a specific area, e.g. a specific model organism. It is therefore difficult for biologists to find information that is not directly related to their field of research. Fortunately, much work was done on linking the various databases by annotating the records with references to external database records. These records and links form a complex graph that takes a lot of time and effort for users to navigate and search. Anybody who has searched the web is familiar with the frustrating experience of pursuing links, backtracking, and returning to the search engine to reformulate the query and begin again. We clearly need to find better ways to search and navigate the biological information in an integrated fashion.