Hi Liam,
Thank you very much for your kind response. I did have done another test with an additional trait state of "no organ". And it had not made big difference. Then I thought about the transition rates among trait states and had difficulty to justify the state of "no organ" with the other states. Thus I was wondering whether there could be any model or test that allow such condition. Indeed the organs were aborted early in development of those three taxa. With current methods, a proper explanation of both prior conditions for these two tests would probably work. Best, Cheng-Chiang ________________________________ From: Liam J. Revell <liam.rev...@umb.edu> Sent: Sunday, July 3, 2016 8:13:14 PM To: Wu, Cheng-Chiang; r-sig-phylo@r-project.org Subject: Re: [R-sig-phylo] Reconstruction of ancestral character state with no organ Hi Cheng-Chiang. Yes, you either need to supply a state for each tip or you can instead supply an ambiguous prior probability distribution. Perhaps it would make sense to add a fourth state that is the absence of the feature? Alternatively, you could imagine that an absent feature might have some trait value if it were observed (like the hair color of a bald man - it exists even if it cannot be observed), but this trait value is unknown. In that case it does make sense to supply a flat prior distribution - then both make.simmap and rerootingMethod can be used to obtain a posterior trait distribution on the unobserved trait's value. All the best, Liam Liam J. Revell, Associate Professor of Biology University of Massachusetts Boston web: http://faculty.umb.edu/liam.revell/ email: liam.rev...@umb.edu blog: http://blog.phytools.org On 7/1/2016 5:28 PM, Wu, Cheng-Chiang wrote: > Dear All, > > I am working on reconstruction of ancestral character state with a trait of > three states (0,1, 2) of a specific organ from 100 taxa. > Per tutorials on > http://blog.phytools.org/2015/09/primer-on-estimating-ancestral-states.html > and http://blog.phytools.org/2013/04/estimating-ancestral-states-when.html, > I set 1/3 as prior probability of being in any of the states for taxa with > the organ but no trait data (missing data). > For the three taxa which don't have this organ, I had to set 1/3 instead of 0 > as prior probability for each state, in order to running rerootingMethod or > make.simmap functions in phytools properly. I tried to set 0 as prior > probability for all states of these three taxa in the matrix, and I got the > following error messages by runing rerooting Method or make.simmap: > > rerootingMethod: > Error in fitMk(tt, yy, model = model, fixedQ = Q, output.liks = > TRUE)$lik.anc[1, : > incorrect number of dimensions > > make.simmap: > Error in 1:nrow(L) : argument of length 0 > > I thought about deleting the three taxa without such organ from my dataset, > but they are critical in the phylogeny. > > May I ask for advice on solution for this issue? > > Thanks, > Cheng-Chiang > > > [[alternative HTML version deleted]] > > _______________________________________________ > R-sig-phylo mailing list - R-sig-phylo@r-project.org > https://stat.ethz.ch/mailman/listinfo/r-sig-phylo > Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/ > [[alternative HTML version deleted]] _______________________________________________ R-sig-phylo mailing list - R-sig-phylo@r-project.org https://stat.ethz.ch/mailman/listinfo/r-sig-phylo Searchable archive at http://www.mail-archive.com/r-sig-phylo@r-project.org/
