Intuiting the Strength of Negative Selection

[davecarlson]

I was going to make some plots, but it seems that somebody has already done it!

(see pages 324-327)

[swamidass (S. Joshua Swamidass)]

What do you make of this graph @davecarlson?

The fixation probability is given for heterozygous, harmful mutations. The vertical line indicates selection coefficients that satisfy Nes = 0.5 and roughly separates the domains where genetic drift (left) and selection (right) dominate the fate of new mutations. The effective population size Ne was estimated from DNA sequence diversity data in the fruitfly Drosophila miranda

https://www.nature.com/scitable/topicpage/negative-selection-1136/

[davecarlson]

This is basically a visual depiction of nearly neutral theory. As the product of the selection coefficient (s) and effective population size (Ne) increases, the probability of a deleterious mutation being fixed decreases. At values greater of Nes >> 1, selection will dominate and the mutation has very little probability of fixing. But at values of Nes < 1 or so, mutations are effectively neutral, and the probability of fixation for the harmful mutation is not much different than the probability of any neutral mutation fixing by drift.

Of course, this sort of analysis neglects all kinds of important information like the local rate of recombination and frequency of background selection and selective sweeps, which can have profound effects on the probability of fixation of mutations (both neutral and deleterious) that are nearby.

[Joe_Felsenstein]

With the formula I gave for fixation probability of new mutants, you should be able to calculate that curve.

[glipsnort (Steve Schaffner)]

Dominance matters, too, at least for the speed at which alleles fix or are eliminated. See this paper: Selective Strolls: Fixation and Extinction in Diploids Are Slower for Weakly Selected Mutations Than for Neutral Ones

[Dan_Eastwood]

My intuition is that larger populations imply stronger (within species) competition, and therefore more intense selection. When resources are limited, any advantage in finding and/or utilizing resources more quickly will add additional stress to the less fit.

In my tinkering with genetic algorithms I wanted to retain variability in the population to avoid getting stuck in local maxima. A strict fitness cutoff tended to get stuck a lot, so I switched to a “three strikes” rule; individuals in some lower percentile of fitness would receive a “strike”, but still have some opportunity to reproduce, getting selected out only after 3 strikes. The percentile cut-off would float to keep a limited population size, rather than the population size being strictly fixed. I don’t know if this is biologically realistic, but my GA converged to good solutions more rapidly.
(I recall this approach could cause a glut of high fitness clones, which had to be cleaned out occasionally.)