Biology Reference
In-Depth Information
(A)
(B)
FIGURE 3.4
Drawing inferences from complex networks through modeling.
experimental evidence that directly relates to the protein
interaction
[84]
, or can include a broader range of bio-
logical relationships. Methods are emerging to score the
likelihood that a given protein pair will physically
interact, given all the other known single-color relation-
ships about that protein pair. Prediction strategies that
model single-color relationships as being independent of
one another have proved worthwhile
[54,127,128]
. More
sophisticated
interdependency
between multiple
interaction
types
further
improve the quality of
interaction predictions
[128
133]
.
With such confidence scores, interactome networks can
be represented as probabilistic networks in which each
edge is assigned a weight representing the posterior prob-
ability that the edge is real. Sampling many 'deterministic'
subgraphs from the probabilistic network measures the
fraction of sampled networks containing a path between
e
approaches
that
explicitly model
the
