Dear all
I isolated an archaeon with 93% similarity to the top-hit strain in 16S rDNA. First, I want to know which method is better to generate a phylogenetic tree, neighbor-joining or maximum likelihood? Second, how many strains should be used for the alignment?
Thanks.
Without a doubt maximum likelihood is better that distance methods (Neighbor joining) in every circumstance.
Distance methods are much, much less robust to departures from model assumptions (biased data) than maximum likelihood methods. So if even after you have found a good model to describe sequence evolution (JC, HKY, GTR with across site rate parameters) maximum likelihood, a model is never perfect so it is very important to have a method that is the least sensitive to the bias that cannot be corrected.
Furthermore, maximum likelihood methods can return a confidence set of trees that are not significantly different from the ML tree whereas distance methods return one tree only giving no sense for alternatives that many be just as good.
- Badges
- Science topic
- Similar topics
- Philosophy
- Esthetics
- Interpretation