Biology Reference
In-Depth Information
Baconian science in which hypothesis-driven research is giving way to
hypothesis-free experiments and data collection.
34
Such work involves
the collection and analysis (“mining”) of large amounts of data in a
mode that is not driven by particular hypotheses that link, say, specifi c
genes to specifi c functions. Genome-wide association studies (GWAS),
for instance, search for genes associated with particular phenotypes by
searching large amounts of genotypic and phenotypic data for correla-
tions that statistically link specifi c genomic loci to particular traits. With
enough data, no prior assumptions are needed about which loci match
which traits. This sort of computer-statistical analysis can generate rela-
tionships between biological elements (pieces of sequence, phenotypes)
based not on shape, or function, or proximity, but rather on statistics.
In other words, the data-statistical approach provides new dimensions
along which biological objects can be related.
The examples in this chapter demonstrate that bioinformatics is di-
rected toward collecting large amounts of data that can subsequently be
used to ask and answer many questions. Thus the terms “data-driven”
and “hypothesis-free” have become focal points of debates about the
legitimacy of bioinformatic techniques and methods. Both biologists
and philosophers have taken up arguments over whether such meth-
ods constitute legitimate ways of producing biological knowledge. The
contrasting positions represent tensions between new and old forms of
knowledge making and between new and old forms of organizing bio-
logical work. Indeed, the sharpness of these epistemological disagree-
ments is further evidence that bioinformatics entails a signifi cant chal-
lenge to older ways of investigating and knowing life.
The biochemist John Allen has been especially outspoken in argu-
ing that there can be no such thing as hypothesis-free biology and even
that such work is a waste of time and money: “I predict that induction
and data-mining, uninformed by ideas, can themselves produce neither
knowledge nor understanding.”
35
Alarmed that hypothesis-driven sci-
ence is portrayed as “small-scale” and “narrowly focused,” critics of the
data-driven approach argue that true science is not possible without a
hypothesis, that a lack of a hypothesis indicates a lack of clear thinking
about a problem, and that the data produced by hypothesis-free science
will be uninterpretable and hence useless.
36
Sydney Brenner, another
particularly outspoken critic, describes data-driven biology as “low
input, high throughput, no output.” In his view, both systems biology
and genome-centered biology work at the wrong level of
abstraction
—
simply examining more data at a genome-wide level will not help biolo-
