I used SLAC (Single Likelyhood Ancestor Counting) in the program HyPhy to detect negativ selected sites in a gene. It looks all nice and plausible but I would like to know how exactly the program calculates the p-values. I read the original paper were they mention using an extended twotailed binominal distribution of the expected number of synonymous substitutions to test the significance of positiv and negativ selection. But they don't specify the statistical test used for this. Maybe it is somehow obvious just not for me... Could it be some kind of Chi-square test?

Maybe someone has the statistical knowledge to answer this. If more information is needed about the tested values, please ask and I will provide it.