Hi Stacey, thank you for your help! Yes, SIMMAP deals with short branches: I used this tree to map characters in Mesquite, and then I kept it as the topology was ok, but finally this was not because of SIMMAP ! :)
For the sampling, if I only have character data for 64 taxa, I thought I could : (1) keep the total 116 species tree for the mapping even if 52 species are no documented, (2) run the total tree, remove the 52 species, and do the mapping on the remaining 64 species (but is it possible to do that easily on 10000 trees??), (3) run the 64 taxa tree and use it for the mapping. i think it is not good for the mapping to have a tree with taxa without information on character states in it. So a reduced tree might be better... the 64 species tree is quite well resolved and the topology is the same than in the total tree, but I am interested in knowing if solution (2) is feasible! What do you think? Cheers, m 2011/3/30 <[email protected]> > Hi Marion, > About your questions, method (1) would still take into account phylogenetic > uncertainty, just among the 64 taxa instead of 116. It seems a shame, > though, > to lose those extra taxa. I'm surprised SIMMAP can't deal with the short > branches. > Second, about the sample from MrBayes, you will want to use the trees from > the > .t file, not the trees from the .trprobs. The ones in the .t file do have > branch lengths. Also you do not want to take the most probable trees but > instead a random sample (a sample from the .t file) for mapping. This is > how > to account for phylogenetic uncertainty appropriately. > Cheers, > Stacey > > > Quoting Marion Chartier <[email protected]>: > > Hi everybody, >> >> I am a PhD student new on the list! I am working for my thesis on the >> evolution of pollination in the Araceae family, and so beginning to >> perform >> phylogenetic analyses! >> >> I am using SIMMAP to map floral traits on a phylogeny of the family, and >> to >> identify correlations between the evolution of pollination type and some >> floral traits. >> >> I have some questions about the parameters to set in the analysis of >> correlation! >> >> >> Method (1) >> >> So far, I have contructed a phylogeny (64 genera), and taken the best ML >> tree topology to map 20 characters in SIMMAP. This was more simple for me, >> as I had to remove taxons from a big 116 genera tree before running the >> SIMMAP analysis. I also did this because the big tree was well supported, >> I >> needed to get branch lenghts different from zero, and the topology was the >> same than in my Bayesian tree. >> >> Thus, I was thinking of largely improving the number of samples >> (mutationnal >> maps simulations) in the SIMMAP parameters of the correlation analysis... >> (like from 10 (default) to 100,000). >> >> >> Method (2) >> >> A better (I guess) way to do it is to take the 10000 most probable trees >> from a Bayesian analysis of my 64 genera alone (less supported than the >> big >> tree), as it is suggested in the manual, and keep a small number of >> samples. >> >> >> So here come my questions: >> >> >> 1- I know that the method (1) doesn't take acount to the phylogenetic >> uncertainty, is it anyway statistically correct? >> >> 2- When running MrBayes, I get an output file containing the most probable >> trees, but with no branch lengths... (the .trprobs file). I guess this are >> the trees I should import in SIMMAP for method (2), but is it ok without >> the >> branch lengths?? >> >> >> Thank you very much for helping me! >> Merci beaucoup de votre aide ! >> >> Marion >> >> -- >> *¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤* >> Marion Chartier >> PhD director: Marc Gibernau >> UMR Ecofog >> Campus Agronomique - BP 709 {|} >> 97 387 Kourou Cedex | >> FRENCH GUIANA |() >> mail : [email protected] || >> *¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤* >> >> [[alternative HTML version deleted]] >> >> >> > > > -- *¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤* Marion Chartier PhD director: Marc Gibernau UMR Ecofog Campus Agronomique - BP 709 {|} 97 387 Kourou Cedex | FRENCH GUIANA |() mail : [email protected] || *¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤*¤* [[alternative HTML version deleted]]
_______________________________________________ R-sig-phylo mailing list [email protected] https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
