Noise propagation in transcription networks

Source: Dunlop et al (2008). Regulatory activity revealed by dynamic correlations in gene expression noise. Nature Genetics 40(12):1493-1498.

Biological events are stochastic in nature. Random fluctuations in protein concentration, expression and etc relays noise through the transcription network via the regulatory links. For example, a random decrease in the concentration of a repressor results in an increase in the expression of its target gene; however, only if the concentration of the repressor falls within an "active" range in which the expression of the target genes is sensitive to small chanages in the repressor content (see Fig. below).
Thus, observed correlations between the expression of different genes may be the result of a direct or indirect regulatory process. However, in addition to intrinsic noise (fluctuations in the expression of a given gene), we should also consider the extrinsic noise in which all the genes are uniformly affected by a given change (e.g. a random increase in the ribosome content of the cell increases the expression of all the genes). Extrinsic noise causes false positive correlation (see Fig. below).
Thus, any measurement of correlations must be normalized by the effect of extrinsic noises. In this paper, the authors use both stochastic modeling and experimental validation to make the case for this phenomenon.