Re: [R-sig-phylo] chronopl
Hi Hilmar, Hilmar Lapp wrote on 11/04/2011 21:54: Hi Emmanuel: On Apr 11, 2011, at 12:55 AM, Emmanuel Paradis wrote: You should not used the transformation suggested on this wiki (it is a bit outdated, NPRS has been removed from ape). In most situations, you should just leave S=1. The wiki can be updated (by anyone, in fact). How would you suggest should this be corrected to be current? Just say that the default value of S=1 should be left as it is. And maybe add a line saying that the help page is the reference for updates. Emmanuel -hilmar -- Emmanuel Paradis IRD, Jakarta, Indonesia http://ape.mpl.ird.fr/ ___ R-sig-phylo mailing list R-sig-phylo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Re: [R-sig-phylo] chronopl
Hi Emmanuel: On Apr 11, 2011, at 12:55 AM, Emmanuel Paradis wrote: You should not used the transformation suggested on this wiki (it is a bit outdated, NPRS has been removed from ape). In most situations, you should just leave S=1. The wiki can be updated (by anyone, in fact). How would you suggest should this be corrected to be current? -hilmar -- === : Hilmar Lapp -:- Durham, NC -:- informatics.nescent.org : ___ R-sig-phylo mailing list R-sig-phylo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
Re: [R-sig-phylo] chronopl
Rafael, You should not used the transformation suggested on this wiki (it is a bit outdated, NPRS has been removed from ape). In most situations, you should just leave S=1. chronopl() is not "broken", but maybe not as reliable than Sanderson's r8s, though I have seen examples where both programmes agree a lot. What is needed is someone who could spend a few days (full-time) on testing and improving chronopl in depth (which I cannot at the moment). Cheers, Emmanuel Rafael Maia wrote on 09/04/2011 04:36: hello list, I have been trying to work with chronopl to obtain molecular dating and ultrametricize a couple trees obtained from Mr. Bayes (at this point, I'm trying to learn what these functions and methods do, the differences between them, and follow examples from published papers). I followed the instructions in the R phylo wiki on my datasets, in order to obtain mean substitution numbers: mytree$edge.length = mytree$edge.length * N where N is the number of sites. After doing this, I followed the cross validation method to obtain the smoothing parameter, which was of 1 for one tree and 10 for the other. However, when I plot the trees, they are identical. In fact, if I use ridiculously small (0.01) or large (100) values of lambda, they are still identical. This does not happen if I use the original edge.length of the tree, but I'm guessing this is probably nonsensical given how the likelihood function is calculated. I have noticed that the version of chronopl I'm using (ape v.2.6-2) has an S argument. if I use the original edge.length and specify S=N, then the results are in fact affected by the value of lambda, and are very different than if I use the modified edge.length as specified above (indeed, the cross-validation result for lambda is also different). Looking online for information, I read that the chronopl function is "broken", not realiable/not working. Is this still true? Am I not supposed to alter the branch lengths as specified in the wiki? Should I use the S argument? Many thanks for any help! Abraços, Rafael Maia --- webpage: http://gozips.uakron.edu/~rm72 "A little learning is a dangerous thing; drink deep, or taste not the Pierian spring." (A. Pope) Graduate Student - Integrated Bioscience University of Akron http://gozips.uakron.edu/~shawkey/ ___ R-sig-phylo mailing list R-sig-phylo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-phylo -- Emmanuel Paradis IRD, Jakarta, Indonesia http://ape.mpl.ird.fr/ ___ R-sig-phylo mailing list R-sig-phylo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-phylo
[R-sig-phylo] chronopl
hello list, I have been trying to work with chronopl to obtain molecular dating and ultrametricize a couple trees obtained from Mr. Bayes (at this point, I'm trying to learn what these functions and methods do, the differences between them, and follow examples from published papers). I followed the instructions in the R phylo wiki on my datasets, in order to obtain mean substitution numbers: mytree$edge.length = mytree$edge.length * N where N is the number of sites. After doing this, I followed the cross validation method to obtain the smoothing parameter, which was of 1 for one tree and 10 for the other. However, when I plot the trees, they are identical. In fact, if I use ridiculously small (0.01) or large (100) values of lambda, they are still identical. This does not happen if I use the original edge.length of the tree, but I'm guessing this is probably nonsensical given how the likelihood function is calculated. I have noticed that the version of chronopl I'm using (ape v.2.6-2) has an S argument. if I use the original edge.length and specify S=N, then the results are in fact affected by the value of lambda, and are very different than if I use the modified edge.length as specified above (indeed, the cross-validation result for lambda is also different). Looking online for information, I read that the chronopl function is "broken", not realiable/not working. Is this still true? Am I not supposed to alter the branch lengths as specified in the wiki? Should I use the S argument? Many thanks for any help! Abraços, Rafael Maia --- webpage: http://gozips.uakron.edu/~rm72 "A little learning is a dangerous thing; drink deep, or taste not the Pierian spring." (A. Pope) Graduate Student - Integrated Bioscience University of Akron http://gozips.uakron.edu/~shawkey/ ___ R-sig-phylo mailing list R-sig-phylo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-phylo