Experimental noise does not account for all of the observed variance of epistasis. (a) Comparison of experimentally measured variance (solid lines; shaded areas: 25 to 75% confidence intervals) and variance caused by experimental noise (dashed lines). If one of the two mutations is neutral, noise accounts for all of the observed variance (black). When deleterious mutations are combined, noise accounts for less than half of the observed variance (red, G01 ≈ -0.7). (b) Ratio between total observed variance and noise-generated variance as a function of the log growth of the knockouts being combined. For deleterious knockouts, the ratio can be significantly greater than 1.
Velenich and Gore Genome Biology 2013 14:R76 doi:10.1186/gb-2013-14-7-r76