The challenges for CP in bioinformatics: a biochemist's perspective
===================================================================

Ludwig Krippahl



Abstract
--------
The application of CP to bioinformatics has been an active field of
research for nearly two decades. And, in this period, CP has proven to
be a useful approach to problems ranging from sequence analysis
through molecular structure to metabolic patwhays. And, yet, the
penetration of CP methods in the bioinformatics community has been
disproportionally modest. One important reason for this seems to be
that the CP community, understandably, is focused mainly on the core
issues of constraint programming, giving a lower priority to other
aspects. Such as those implementation details that, though less
innovative or meaningful in themselves, are nevertheless essential for
any computational tool to be used. Or hybrid methods that complement
the strengths, and overcome the weaknesses, of CP by combining it with
other approaches, trading theoretical purity for functionality. In
short, while CP researchers are naturally concerned mainly with CP
research, biochemists are interested in bioinformatics as a tool for
their own purposes. This gap is being bridged, but only slowly.

This talk will present a brief overview of some achievements of CP in
bioinformatics, proposing that high expectations for CP in this field
are not unjustified, and give two examples that give an indication of
what factors are most important for building CP tools based
biochemists actually use. One is the application of CP to protein
docking, and which was already used to model several protein
complexes. The other example, far less successful, is the application
of CP to processing of structural information and why it is still
short of being useful despite the advantages that CP provides in this
case.