If d(R, T1, T2) is negative, it would mean that the expression relationship between R and T1 can be better explained when the expression of T2 is known, and thus T1 and T2 are not independently regulated by R, but they affect each other possibly due to buffering.
Studies in yeast and bacterial systems have shown that gene expression noise decreases as mean expression levels increase, a relationship that is controlled by promoter DNA sequence.
An inverse expression relationship means that when the expression level of the miRNA is high (up-regulated), the target mRNA should be down-regulated based on the principle that miRNAs deregulate the expression of targeted mRNAs [ 10].
Studies in yeast and bacteria have shown that basal promoters are important determinants of the inverse relationship between mean expression and noise (Tirosh and Barkai 2008; Hornung et al. 2012; Carey et al. 2013).
Alemu's measure of expression variability (EV) tries to account for the above described relationship between mean expression and variability in a distinct way and to provide a measure of variability which is independent of expression mean (Additional file 3: Figure S2).
These thresholds were chosen based upon visual inspection of plots evaluating the relationship between mean expression level and fold change (commonly known as 'MA plots' in the context of microarrays), attempting to balance the tradeoff between maximal sample size and reduced noise.
> -wrap-foot> > Based on the above observations, one may suspect that the observed changes in expression variability are simply artifacts of some non-linear relationships between mean expression level and variability.
The second reason is that most previous studies have focused on the preservation of mean expression levels, as opposed to the preservation of co-expression relationships.
As far as we are aware, this is the first study that establishes relationships between inter-sample expression variability, mean expression level and coexpression.
Full results of the analysis are shown in Additional file 2. Figure 2 shows the relationship between log2 fold change in expression between Ribo-Zero and poly(A) and mean expression level across all samples.
These results imply that distal CRMs do not affect the noise mean expression relationship and are consistent with studies in yeast implicating promoter sequence and nucleosome occupancy at the transcriptional start site in expression noise control (Tirosh and Barkai 2008; Hornung et al. 2012; Carey et al. 2013).
