Information Technology Reference
In-Depth Information
sequence of the correspondent gene on DNA) it is much closer to the genotype
pole (it is produced by the simple application of genetic code to the DNA
sequence, setting aside for a moment all the complications of RNA editing,
splicing and so forth.).
The classification of organisms on the basis of their phenotypic similarities is
thought to be related with the classification based on corresponding genotypic
make-up. When we discover a genotype/phenotype relation in terms of
consistency of two classifications respectively based on a phenotypic and a
genotypic character, we have the proof of a functional relation between the two
features that somehow 'sensed' the evolutive history in the same way giving rise
to congruent similarity spaces for the studied organisms.
In this application, we investigated the classification of organisms based on
similarities as for metabolic wiring (phenotypic feature) in comparison with the
classification of the same organisms based on the sequence similarities of the
shared metabolic enzymes. The finding of a strong correlation between the two
allowed us to both give a strong proof of 'metabolic wiring' as a specific
phenotypic character under evolution constraints and to confirm, by means of a
completely independent analysis, the crucial role exerted by non-hub connectors
in metabolic networks.
The first computational problem to solve was finding a useful metrics for
network comparison so to have a reliable measure of how much two wiring
architectures differ from each other. A network of
n
nodes can be represented as
a binary square adjacency matrix
n
x
n
where the entry at (
i, j
) position is 1/0 if
there is/there is not an edge from vertex
i
to vertex
j
(see Fig. 8.5).
Fig. 8.5. The network on the left is represented as an adjacency matrix, in which an element in
i,j
position is 1/0 if an arc exists/doesn't exist connecting node
i
to
j
.
