It's possible that Mesquite (and within R) that discrete traits may be interpreted by the software as continuous, depending on how they're stored -- so it could do a reconstruction of a character with tip states of 0, 1, and 2 and get ancestral values of 0.1, 1.75, etc. using squared change parsimony or the equivalent. This isn't something one should do (for one thing, if the trait is unordered, the ancestor of a clade with taxa in state 0 and 2 having a state of 1 doesn't make sense) but it is possible someone could run squared change parsimony on discrete traits if the software mistakenly treats them as continuous.
Best, Brian _______________________________________________________________________ Brian O'Meara, http://www.brianomeara.info, especially Calendar <http://brianomeara.info/calendars/omeara/>, CV <http://brianomeara.info/cv/>, and Feedback <http://brianomeara.info/teaching/feedback/> Associate Professor, Dept. of Ecology & Evolutionary Biology, UT Knoxville Associate Head, Dept. of Ecology & Evolutionary Biology, UT Knoxville On Thu, Jan 31, 2019 at 2:53 PM Liam Revell <liam.rev...@umb.edu> wrote: > Roland. > > You have not done 'squared change parsimony' in that case, but just > parsimony. The problem (among various) is that for a discrete character > there is not always a single most parsimonious set of ancestral states > and there are often many equally parsimonious reconstructions. That's > why some scientists have developed criteria (such as 'ACCTRAN' and > 'DELTRAN') for consistently selecting one or two among the usually many > equally parsimonious reconstructions. One should not be deceived by > this, however. Just because the ACCTRAN and DELTRAN criteria exist > doesn't mean that these MP solutions for ancestral states are any more > probable than any other equally parsimonious solution (and under some > circumstances all parsimonious solutions can be misleading). > > I believe parsimony ancestral state reconstruction is implemented in the > phangorn function ancestral.pars. > > All the best, Liam > > Liam J. Revell > Associate Professor, University of Massachusetts Boston > Profesor Asistente, Universidad Católica de la Ssma Concepción > web: http://faculty.umb.edu/liam.revell/, http://www.phytools.org > > On 1/31/2019 2:18 PM, Roland Sookias wrote: > > Dear Liam, > > > > Thanks very much indeed - a really nice blog post. However, the issue is > > that I am using discrete states, which are not ordered. (and to make > things > > more exciting there are also polymorphisms and missing data - although > they > > can be worked around I guess) > > > > Is there any way to get ML to yield the same as SCP would (in Mesquite at > > least) for this kind of data? > > > > Best wishes > > > > Roland > > > > On Thu, Jan 31, 2019 at 8:03 PM Liam Revell <liam.rev...@umb.edu> wrote: > > > >> Dear Roland. > >> > >> I didn't really solve your problem of reading a Mesquite result into R; > >> however I just posted about the relationship between unweighted & > >> weighted square-change-parsimony & ML ancestral character estimation on > >> my blog: > >> http://blog.phytools.org/2019/01/unweighted-weighted-square-change.html > . > >> (In summary, you should be able to do what you want within R.) > >> > >> All the best, Liam > >> > >> Liam J. Revell > >> Associate Professor, University of Massachusetts Boston > >> Profesor Asistente, Universidad Católica de la Ssma Concepción > >> web: http://faculty.umb.edu/liam.revell/, http://www.phytools.org > >> > >> On 1/31/2019 12:48 PM, Roland Sookias wrote: > >>> Dear all, > >>> > >>> I want to get squared change parsimony ancestral state reconstructions > >> for > >>> a matrix into R. > >>> > >>> I have tried the asr_max_parsimony() function from castor, but the > >> results > >>> are not the same as in Mesquite for example. To take an example, a > >>> two-taxon clade with a unique state for these taxa optimized as having > a > >>> different state (shared by near relatives) at the ancestral node. This > >>> doesn't happen in Mesquite and is *not* what I want. > >>> > >>> If there were some way to get the results from Mesquite into R that > would > >>> also be OK, but I cannot figure out the node numbering system in > >> Mesquite, > >>> to match them up with nodes as numbered by ape. > >>> > >>> Any help would be *very* greatly appreciated. > >>> > >>> Best wishes, > >>> > >>> Roland > >>> > >>> [[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/ > > > _______________________________________________ > 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/
