Biology Reference
In-Depth Information
handling and modeling. In the remainder, we will therefore concentrate on the
top-down approach and discuss some questions pertaining to the epistemological,
conceptual, and ontological foundations of this young biological discipline. We
do not aspire to solve these questions here.
8.1. Decomposition of large networks and functionality
The dynamics of a large network is neither explainable from an account of its
components alone, nor does an account of the overall dynamics of a network
count as its own explanation. In contrast, an explanation may refer to the roles
of subnetworks or modules in the dynamics of the whole network. An important
step in systems biological top-down modeling is, therefore, the decomposition
of a system into interacting subsystems. As already mentioned, there are two
modularization strategies, one using criteria of functionality to identify mod-
ules and the other proceeding by a mathematical analysis of network structure
(Section 6.2). The first strategy to decompose a network starts from a biological
capacity that one can attribute to a network, e.g., regulation of a developmental
pathway or processing of some signal. This capacity may be analyzed further
by singling out the functional contributions of sub-networks to the capacity.
A sub-network to which a function can be ascribed, then, is considered a func-
tional module. (Within the module, functionality need not be, and usually is not,
localizable at any particular component but can be distributed.) The concept of
function, however, is controversial within philosophy of science, and different
accounts compete with each other (for a review, see, e.g., Wouters, 2005). We
therefore must specify which notion of function is applicable here. An approach
that fits the strategy described well is Cummins' functional analysis (see Boogerd
et al., 2005) according to which the function of a system's component is its sys-
temic role in contributing to a capacity of that system (Cummins, 1975).
11
Since
most functions are not performed permanently, 'role' is to be understood dispo-
sitionally here. Functional analysis may be based on the results of physiological
experiments, as in the biochemical analysis of metabolic and signaling pathways
(whose results may be further used for bottom-up pathway modeling), or, to
some extent, be performed mathematically, like in the analysis of input-output
data, as long as the results can in some way be related to structural data.
12
In
11
Other accounts of function refer to the evolutionary history of a biological trait and single out the role a
trait was adapted for as its proper function. These accounts are barely applicable to systems biology, which
mostly lacks an evolutionary perspective and in most cases investigates a network as it is now without taking
into account its history (cf. also note 16).
12
One point should be added to our rendering of Cummins's account: Functional analysis not only refers to
a system and its capacities, but also includes a description ('analytic account') of this capacity. Therefore,
the capacity and functions by which the system's components contribute to it cannot be read from the system
alone, but are always relative to the structuring of the capacity by the description given by scientists. Functions
