I'm using mlcoalsim in order to compare the observed distribution of four summary statistics (neutrality tests) with those obtained from nine different demographical models. Nevertheless, the probability that simulated values be smaller than or equal to the observed values is the same for all nine models (including the standar coalescent model). I suspect that my models are not well inputed (for instance, I'm not sure how to transform times from years to 4N generations. I assumed that t =r/4N; where t is the time in 4N generations and r is the number of generations) or they are not really different. I would be grateful If someone could help me and give me some advices to make a succesfull simulation with mlcoalsim (feel free to write me in Spanish or English).
Thank you in advance.
I first need to know what is 'N' in the equation t = r/4N?
Usually, N means the number of gene copies. For autosomic genes in diploid organisms it's corrected by the number 4 as a multiplier and then N becomes your number of samples. If you are working with haploid organisms, then N is corrected being multiplied by 2, so t=r/2N.
In any case, the problem with models is not relative to the equation it is simple to solve if you have the number of generation estimated for some cladogenetic event. My major concern is about you have the same likelihood rate for all models.
It's difficult to know just with the description. I recommend you to try some example data, make a null distribution of your data or use the process you're using to get your data with some other dataset you already know the result. Another alternative is to try to deal with different software (if it's possible). Sorry for being so broad in my comment but there is a lot of possibilities for your issue.
Good luck.
Best!
Remember always that you are modeling phylogenetic relationships, not evolution. Evolution is not necessarily dichotomous as in hierarchic cluster analysis (of which cladistics is a variant) or Markov chains. Cladistics is also blind to ancestor-descendant relationships. The optimal model in cladistics is the most probable or most credible generation of a dichotomous tree that is not found in nature.
Thank you so much for your sugestions. After all this time (actually on november 2017) I noticed that I had been gotten equal likelihood values for all coalescent models because the software does not make any kind of "warning advice" even if you are using input files that are wrong (or incomplete). I mean, It just run assuming a standar coalescent model (and not take into account demographic and time parameters).
- Badges
- Science topic
- Similar topics
- Biological Science
- Evolutionary Biology