Abstract

It is now abundantly clear that we will have access to thousands of
sequenced genomes within just a few years. The question "How should the
effort to interpret this wealth of data be structured?" may be a bit
disconnected from the work of most biologists.  Perhaps it should be cast
instead as "How can this accumulation of data be used to characterize or
locate specific genes of interest?"  It is becoming clear that evidence
provided by a growing set of techniques can be integrated to support
characterization of genes.  The most effective techniques to date have been

        1. analysis of clustering on the chromosome,
        2. inference of function based on detection of gene fusions,
        3. use of gene-occurrence profiles, and
        4. analysis of regulons based on analysis of putative operator sites.

In my talk I will discuss instances in which these techniques have been used
to make predictions that have been later confirmed in the wet lab.  I will
advance the position that we can now produce a rapidly growing set of
conjectures that are reasonably reliable, and that this fact will rapidly
elevate the role of bioinformatics in our efforts to understand organisms.