Biology Reference
In-Depth Information
in bioinformatics: setting the data free to tell their own story. Because
such data are not understood to be tied to a specifi c hypothesis, they
may have many uses, many voices, many stories to tell. This is a funda-
mentally different mode of investigation that is increasingly central to
biological work.
44
Franklin-Hall has also suggested that part of what is going on with
exploratory science is a maximization of effi ciency in scientifi c inquiry.
Where a narrow instrument dictates that only one or two measurements
can be made at once, it pays to think very carefully about which mea-
surements to make. In other words, it is effi cient to use the instrument to
test a very specifi c hypothesis. With a wide instrument, however, this re-
striction is removed, and effi ciency of discovery may well be maximized
by making as many measurements as possible and seeing what interest-
ing results come out the other end. However, this economy of discovery
can be realized only if the investigator has some means of coping with
(processing, analyzing, storing, sharing), the vast amount of data that
are generated by the instrument. This is where the computer becomes the
crucial tool: effi ciency is a product of bioinformatic statistical and data
management techniques. It is the computer that must reduce instrument
output to comprehensible and meaningful forms. The epistemological
shift associated with data-driven biology is linked to a technological
shift associated with the widespread use of computers.
Natural philosophers and biologists have been collecting and accu-
mulating things (objects, specimens, observations, measurements) for a
long time. In many cases, these things have been tabulated, organized,
and summarized in order to make natural knowledge.
45
But specimens,
material objects, and even paper records must be treated very differently
than data. Data properly belong to computers—and within computers
they obey different rules, inhabit different sorts of structures, are subject
to different sorts of constraints, and can enter into different kinds of re-
lationships.
46
Bioinformatics is not just about
more
data, but about the
emergence of data into biology in tandem with a specifi c set of practices
of computerized management and statistical analysis. These computer-
statistical tools—expectation maximization, Bayesian inference, Monte
Carlo methods, dynamic programming, genetic algorithms—were ex-
actly the ideas that began to be taught in the new bioinformatics courses
in the 1990s. These were the tools that allowed biological knowledge
to be built out of sequence. These were the tools that have allowed bi-
ologists to reorient toward general, large-scale questions. Rather than
studying individual genes or individual hypotheses one at a time, bio-
