Measuring gene interactions.

Measurement is the assignment of numbers to reality, and valid measurement requires that these numbers represent relevant aspects of reality. I discuss epistatic gene interactions from a measurement-theoretical perspective and argue that the standard measurements of epistasis in classical quantitative genetics have failed to capture aspects of epistasis that are relevant to selection dynamics and adaptation. Instead, the use of statistically motivated measurements such as epistatic variance components has led to the misconception that epistasis is dynamically inert. Here, I review work showing that patterns of epistasis may have profound effects on evolutionary dynamics and discuss how these patterns can be measured.