Biology Reference
In-Depth Information
especially the
in vivo
techniques. Second, many biologically relevant
interactions (as, for example, revealed by genetic interaction studies) are
not direct either — genes can show a strong functional linkage, but still
their products might be located in different parts of the cell or expressed
at different times. Third, the distinction between direct physical binding
and indirect functional effects can be somewhat artificial: two proteins
may interact over only a small segment of their length, or only very
briefly (such as when one protein modifies another posttranscriptionally);
also, proteins may come into contact simply by chance because they are
secluded in the same small intracellular compartment, for example, but it
might appear like they are interacting specifically. Last, making the dif-
ference between a specific interaction and a nonspecific “infrastructure”
interaction is far from obvious: one protein may transport another to a
certain place in the cell in a very specific, important, and tightly regulated
interaction, or merely as part of a ubiquitous and nonspecific service.
Therefore, protein-protein association networks can be said to represent
the smallest common denominator of partnership between proteins:
whenever two proteins form a specific functional partnership, they can be
thought of as being “associated”, independently of what the actual
mechanism of their association is.
Notably, association is also a very useful concept for another reason:
the definition matches the results of certain types of prediction algo-
rithms. Not all computational interaction prediction algorithms predict a
direct physical interaction between two proteins; instead, some algo-
rithms merely predict an association. This is particularly true for algo-
rithms that are based on the effects of evolutionary selection. Such
predictors work by detecting deviations from a random assortment of
genes in multiple genomes. Simply put, a pair of genes — if functionally
not related — should evolve more or less independently. Both genes
should freely move around within a genome, for example, when chro-
mosomes are broken up and reassembled over time; and the two genes
should independently undergo gene duplications, gene losses, gene
transfers, and other such events. Any deviation from this random expec-
tation, if detected in a sufficiently large number of genomes, can be taken
as evidence that selection has somehow acted similarly on both genes.
This in turn means that the two genes somehow share certain aspects of
