Biomedical Engineering Reference
In-Depth Information
contrast these with the
parameters
of robustness—those mechanisms by which
these principles are realized. The principles and parameters metaphor is derived
from linguistics (7), where the principles are defined as the invariant properties
of universal grammar and the parameters the local rules and practices of lan-
guage. Here we extend these principles to include feedback, modularity, spatial
compartmentalization, distributed processing, and the extended phenotype. An-
other way of thinking about the principles is as higher grades in a theoretical
taxonomy of robustness. All mechanisms employing some form of redundancy
are classed together, as are those employing modularity and so on. As we work
down the classificatory tree of robustness, we eventually hit the unique me-
chanical instantiation that gives rise to robustness. Our classification is more
Linnean than Darwinian, as we have no external principle with which to organ-
ize mechanism.
We give a brief introduction to each of these principles below, and subse-
quently go on to discuss in more detail a few models developed to address spe-
cific robustness mechanisms in biology.
3.1. Redundancy
A common means of identifying the function of a gene is to perform a
knockout experiment, removing or silencing a gene early in development. By
assaying the resultant phenotype, the putative function of the absent gene can be
inferred. In many such experiments there is no scoreable phenotype: the knock-
out leaves the phenotype in the wild-type condition. Biologists refer to a gene
x
on a background
y
as functionally redundant (57). This is taken to mean that the
target gene is one of at least two or more genes contributing to the phenotype
epistatically (27). Removal of a redundant gene
x
leads to compensation by re-
maining members of a redundant set
y
. Let
f
(
g
) be the fitness of gene or genome
g
; then redundancy implies that
f
(
x
,
y
) =
f
(
y
). When
y
has a cardinality of one
and
y
=
x
, then functional redundancy reduces to the special case of a redundant
copy of
x
. Redundancy as a principle is more general, and describes any case in
which the mechanism of robustness is only operative upon perturbation. Hence
redundancy is a variational property, not contributing to fitness directly, but in-
directly operating at the population level. Individuals with a redundancy prop-
erty are not fitter than those without, but those without will on occasion suffer
the consequences.
True redundancy might be rarer than "artefactual" redundancy, or experi-
mental neutrality, in which the effect of perturbation remains below an experi-
mental detection limit (47). Assuming that we are able to detect small changes,
the degree of redundancy describes the degree of correlation among genes con-
tributing to a single function. Models of redundancy in biology tend to focus on
the evolutionary preservation of redundant components, and hence employ
