This version of a is epistatic to b and c: it hides their output.
This version of c is epistatic to a and b: it hides their output.
Epistasis typically applies to a certain allele, or version, of a gene. Epistasis depends on how the protein the allele codes for actually functions. In our analogy, epistasis depends on what the workers do in our process.
Now we’ll add epistasis to our exaple. Let’s say a version (or allele) of a is broken so that it contains no instructions. Worker A wouldn’t be able to put paint into the tray, and we would end up with a blank poster—even though workers B and C are still doing their jobs.
This broken version of a is epistatic to b and c: the final product (a blank poster) shows no evidence of what B and C have been told to do. We can’t tell if B’s instructions said to add red or blue, or if C’s said to draw a circle or a square.
The important aspect of epistasis is that it doesn’t just influence the phenotype, it hides the output of another gene or genes.
Now let’s imagine that a is working, but c is broken. This version, or “allele,” of c is epistatic to a and b: the output shows no evidence of what A and B are doing. Even though the output is again a blank poster, the cause is different than when a was broken.
Because of its role in the process, no allele of b can be epistatic to a or c. Changing the color of the dye, or even adding no dye at all, cannot hide what workers A and C are doing.