QUESTION 1: What assumptions does the Wright-Fisher model make?
The wright fisher model assumes a population of randomly mating hermaphrodites with a constant size. In this version, we are also assuming that no new mutations occur. The assumption of constant size is probably the worst assumption we are making (this assumes that the yeast are dying at the same rate they are dividing).
The code I used for the the Wright-Fisher model is here. This code is
for the Bio5488 allele as the dominant allele. It is easy to see how to change it to make the Bio5488
allele recessive.
You can also find updated code using a standard probability "trick" called roulette wheel selection here and also the potential to handle co-dominance and other dominance issues in a common framework here
A student in 2008 actually wrote a really cool answer as can be found here
Question 2: What fitness value gives 95% fixation of the Bio5488 allele, is this what you expected (use max generation=1000)?
At a fitness value of 3, the bio5488 allele usually fixes at or above 95%. This may seem like a small fitness to fix 95% of the time, but remember that this means you have three times the amount of offspring as your wt neighbor (that's pretty fit!).
Question 3: Now do the same for Bio5488 as a recessive allele. What fitness value gives 95% fixation of the Bio5488 allele? What happend and why?
It doesn't matter how high the fitness value for homozygous Bio5488 is, it will never fix 95% of the time. The most it will fix is between 10 - 20%. This is because in the first generation the het has no selective advantage, so the chance of obtaining an individual that is homozygous for the Bio5488 allele is very small (1/40000 to be exact). If you do get an individual that is homozygous for the Bio5488 allele it will almost definitely fix. So when you give the homo. Bio5488 genotype a fitness of 1000000000 you are seeing the probability of getting a homozygote for the Bio5488 allele before the allele becomes extinct.