Biology Reference
In-Depth Information
mensurate with the potential impact they could have on science
and society . . . the problem could rapidly grow worse.
41
Such warnings have been repeated over and over again by biologists
who believe that it's time to stop accumulating data and time to start
analyzing the data that already exist. Insights into the workings of biol-
ogy, and potential medical or pharmaceutical breakthroughs that might
emerge from this knowledge, are being missed, the argument goes,
because they are buried in the fl ood of data. The data, like an ocean,
constitute an undifferentiated mass that cannot be tamed into knowl-
edge. But this dilemma applies only to older forms of biological work
in which data were to be analyzed one by one. Bioinformatic modes of
work
can
tame vast amounts of data—in fact, for some forms of bioin-
formatics, the more data the better.
The philosopher of science Laura Franklin-Hall has drawn a use-
ful distinction between experiments using “narrow” instruments, which
make a handful of measurements in order to test specifi c hypotheses,
and “wide” instruments, which make thousands or even hundreds of
thousands of measurements. Whereas a Northern blot might be used to
measure the presence and abundance of a couple of species of mRNA,
a DNA microarray is capable of measuring the levels of thousands of
mRNAs at once.
42
The microarray's inventors advocate just this type of
novel use for their instrument:
Exploration means looking around, observing, describing, and
mapping undiscovered territory, not testing theories or models.
The goal is to discover things we neither knew nor expected,
and to see relationships and connections among the elements,
whether previously suspected or not. It follows that this process
is not driven by hypothesis and should be as model-independent
as possible . . . the ultimate goal is to convert data into informa-
tion and then information into knowledge. Knowledge discov-
ery by exploratory data analysis is an approach in which the
data “speak for themselves” after a statistical or visualization
procedure is performed.
43
This notion of letting the data “speak for themselves” is no doubt a
problematic one: all kinds of models, assumptions, and hypotheses nec-
essarily intervene between a measurement and a certifi ed scientifi c fact.
Yet it accurately portrays what many biologists think they are doing
