BIO In the literature: partially resolved

Why diminishing-returns epistasis is so common

In plain words

In many experiments a beneficial mutation helps less in a fitter genome, as if all mutations interact through one shared fitness value. Whether this pattern emerges generically from many hidden molecular interactions, and when, is not settled.

Precise statement

Global epistasis: the effect s of a mutation depends approximately linearly on background fitness F, $s \sim a - b F \text{ with } b > 0$, across many yeast and E. coli data sets. Determine the conditions on a microscopic genotype-to-fitness map (random additive plus pairwise interactions, nonlinear phenotype-to-fitness maps) under which this linear dependence holds for most mutations, and predict b from measurable quantities.

What would settle it

A theory predicting the distribution of slopes b from independently measured interaction statistics, confirmed on combinatorial mutant data.

Status in the literature

A 2021 model (Reddy and Desai) showed that many idiosyncratic interactions generically produce global epistasis; conditions and slope values remain open.