Illustration of binary predictors of noise and plasticity, taken in isolation. The presence of a TATA box strongly predicts noise and plasticity. Homo-oligomerization does not, in isolation, predict plasticity, and its effect on noise is only marginally statistically significant (p = 0.0496). However, these effects become significant when confounding factors are accounted for (Tables 1 and 2). Essentiality predicts noise but not plasticity. To better assess effect sizes using more intuitive noise and plasticity measures, back transformations were performed to restore original units. The mean plasticity residual was added to the mean Box-Cox transformed plasticity score, and then the Box-Cox transform was reversed, so that plasticity corresponds simply to the estimated number of experiments for which expression varies. The noise axis corresponds to the DM metric of Newman et al. . Error bars correspond to 95% confidence intervals.
Brettner and Masel BMC Systems Biology 2012 6:128 doi:10.1186/1752-0509-6-128