Papers in computational evolutionary biology

Archive for April 2011

Epistasis and multi-peaked fitness landscapes

leave a comment »

Epistasis is the dependence of the fitness effect of a gene on other genes. It is thought to be a ubiquitous phenomenon: there is no reason to expect that a single gene has a single, clearly defined function which it can perform regardless of the genetic background it is a part of. There are different kinds of epsitasis. Here, the authors consider the reciprocal sign epistasis, a situation where a mutation of one locus can be either deleterious or adaptive depending on another locus, and the same is true of the latter locus as well, with the former now controlling the fitness effect of mutation.

The result reported in the paper is that reciprocal sign epistasis is a feature of any fitness landscape with two or more peaks. The argument (the proof, in fact) is devastatingly simple: find a path between two peaks, and consider the two mutations leading to and from its fitness minumium. If flipping the order of these mutations preserves the location of the minimum, the two loci involved exhibit reciprocal sign epistasis by definition; otherwise proceed to the new miniumum and do the same thing. This procedure necessarily terminates, because every new minimum has a higher fitness value than the previous, and yet is bounded by the fitness of the lower of the two peaks.

The authors proceed to argue that no such local property can characterise (i.e. form a necessary and sufficient criterion) multi-peaked fitness landscapes. The paper is short, well-written, and contains virtually no math. It is good to see that such simpleyet profound insights are still out there to be had.

Poelwijk, F., Tănase-Nicola, S., Kiviet, D., & Tans, S. (2011). Reciprocal sign epistasis is a necessary condition for multi-peaked fitness landscapes Journal of Theoretical Biology, 272 (1), 141-144 DOI: 10.1016/j.jtbi.2010.12.015


Written by evopapers

April 14, 2011 at 09:54

Posted in Uncategorized

Tagged with , , , , ,