Information Technology Reference
In-Depth Information
This artificial enzyme consists of an activity, a set of specificities, and a
shape. Of these, activity and specificity are defined in the enzyme's gene. Shape
is calculated from activity and specificity when the enzyme is expressed. An en-
zyme's activity is either a function, a terminal, or an output. Enzymes with termi-
nal activities are called glands. Those with output activities are called receptors.
An enzyme has at least as many specificities as its activity has inputs. Each
input is associated with one specificity, and this specificity determines which
other enzyme this input will receive output from. In biological terms, a speci-
ficity determines which enzyme the input will recognize and bind as a substrate.
Furthermore, specificity is defined in terms of shape, specifying the shape of
the enzyme it would ideally bind. Note that, unlike in biology, specificity is
given directly upon other enzymes rather than indirectly upon their products.
This encourages product-substrate linkage to evolve, but without the overhead
of maintaining data as a separate entity during execution.
An enzyme's shape determines how it is seen by other enzymes. A shape
is a pattern that both identifies an enzyme and describes its role within the
system. This latter role compares to biology, where shape and function are also
related; though in biology function is determined by shape whereas here, con-
versely, shape is determined by function. A shape is a vector within a unitary
n
-dimensional space. However, before its derivation can be discussed, it is
first necessary to understand how a collection of enzymes interact to form an
executional structure.
Development
Development is the process that maps from an enzyme system's genome to the
executional structure that it represents. In enzyme GP, an executional structure
develops from the interactions of the components described within the genome.
The interactions that occur are decided by a deterministic constraint satisfaction
process, described below. From a biological perspective, this process resembles
the minimization of binding energies in a system of protein-binding proteins
and attempts to capture a generalized model of the construction of metabolic
pathways within a cell.
To generate a valid executional structure, the hard constraints of development
are that the structure contains a full set of outputs (receptors) and that each
component (enzyme) which occurs within the structure has a full set of input
sources. Further hard constraints may be given by the problem domain. For
example, combinational logic design, the domain used to evaluate enzyme GP,
requires executional structures be non-recursive. Given that the hard constraints
are met, the aim of development is then to satisfy the input preferences of
each enzyme of the executional structure; so that a component's actual input
sources are as close as possible to the preferred input sources described by its
specificities.
