Biology Reference
In-Depth Information
els formulated to integrate new data with previous observations from
related experiments. As the observations become more complex, so too
do the models the biologists must construct to make sense of their data.
And as the models become more complex, the computer becomes an
increasingly indispensable partner in their representation, analysis, and
interpretation.”
92
This description might apply equally well to biologi-
cal databases as to the type of computer models that Keller describes.
The structures and categories that databases impose are models for in-
tegrating and making sense of large sets of data. As categorizations of
organisms, sequences, genes, transposable elements, exon junctions, and
so forth, databases are built on sets of structures or principles of bio-
logical organization that are then tested in experiments. Far from being
lists or collections of information, biological databases entail testable
theories of how biological entities function and fi t together.
This understanding of biological databases as models also demon-
strates that the fl ow and ordering of data are central to the constitu-
tion of biological objects and knowledge in bioinformatics. Here we
have once again followed the data into the structures and spaces inside
computers. Databases, which summarize, integrate, and synthesize vast
amounts of heterogeneous information, are the key tools that allow
biologists to ask questions that pertain to large numbers of sequences,
genes, organisms, species, and so on. Databases allow these objects to
be constituted “out of sequence”—that is, brought into new orderings
or relationships with one another that do not necessarily refl ect their
order in cells or on chromosomes. The form of such relationships is con-
strained, however—fl at fi les and relational databases were not designed
for biology, but rather have their own particular histories. The ways in
which biological objects are related to one another have been condi-
tioned by the structural possibilities and limitations of existing database
models—that is, by the histories of databases themselves.
