Open ben18785 opened 5 years ago
Tried with s=+-1, but the effect is weak relative to neutral. Trying now 10.
Even with s=+-10, the effect is still weak. Investigate.
Now works. The only thing that looks strange is that the neutral figure does not recapitulate Bentley Fig. 2 exactly. The proportions are much larger! However, Bentley must be wrongly captioned since 250 individuals over 1000 generations means there are 250,000 people in total. Hence, there is no way to get 10^-8 as a proportion.
For each of N offspring:
With probability (1 - mu), choose a parent at random from the current population but with a probability of choosing each weighted by its variant's frequency in the population; with probability mu, produce a mutant.
The weighting function should probably be propto (1 + s * f) for each parent.
Note that McVean is not frequency-dependent.