Databases and Systems

for wisdom. Where algorithms tend to get implemented, packaged and shared as
black boxes, systems get built over and over in different establishments, and people
facing the same problems for the first time are always peppering their more
experienced counterparts with questions like “did you use approach/product/standard
X in your project? Was it any good? How do you keep the data up to date? How do
you enforce content quality?” This book was conceived as a resource for people
asking such questions, whose numbers seem to be doubling every six months at
present. It is not possible to give definite answers to these sorts of questions – yes
you should or no you shouldn’t use ACEDB or CORBA or OPM or whatever. The
software changes from month to month, the problems change, the options change.
Today’s technological rising star may be tomorrow’s horror story, and vice versa.
Even the insights derived from bitter experience can be questionable – did you really
diagnose the problem correctly, and if you build the next system the way you now
think you should have built the last one, will the result really be better or will you
simply encounter a different, equally painful set of tradeoffs? Nonetheless experience
is better than no experience, so I asked the contributors to this volume to include in
their articles lessons learned from developing their systems, and to write down their
thoughts on how they might do things differently -- or similarly -- if they were doing
it again.
The contributors represent what I hope will be an interesting, albeit
incomplete, sample of some of the most exciting work being done in bioinformatics
today. The articles are somewhat arbitrarily divided into two sections: Databases and
Software. The intent was that articles that focused more on content would fall into the
former category, while articles that focused on technology would fall into the latter,
but there were a number of borderline cases. My hope is that this collection will be of
interest to readers who have arrived at the interdisciplinary world of bioinformatics
either from the biology side or the computational side (as well as those more distant
migrants from literature, history, business, etc.). The database articles may be more
intelligible to readers with more of a biology background, and the software articles to
readers with more software engineering; hopefully there is something to interest (and
confuse) just about everyone.
The articles in the Database section represent some of the established (or in
some cases recently disestablished!) citizens of the database world, as well some
promising new efforts. The first few articles describe systems focused on the
molecular level; these are mostly multispecies, comparative systems. Karl Sirotkin of
NCBI describes some of the software underpinnings of Entrez, the most widely used
molecular biology resource. The article on HOVERGEN by Duret et al describes an
interesting new approach to integrating phylogenetic and coding sequence data into
an organized whole. Several articles focus on the fast-developing area of metabolic
and regulatory pathway databases, including those by Overbeek et al on WIT,
Kanehisa on KEGG, and Karp and Riley on EcoCyc.
The remaining articles in this section describe primarily databases organized
around organisms rather than molecules. Alan Scott describes the extensive
literature-based curation process used to maintain the high quality standards of
OMIM, the fundamental resource on human genetic disease. My own article looks at
3