Thank You, dear Martin, I wasn't aware about TreeDist, going to explore. :-) Yes, RF has some limits, e.g. single difference close to root... Sincerely, V.
Dne úterý 7. března 2023 12:23:07 CET, Martin R. Smith napsal(a): > Dear Vojtěch, > The performance of calculating Robinson-Foulds topological distances > can be improved using the algorithm of Day (1985), which is implemented > in the R package TreeDist: > https://ms609.github.io/TreeDist/reference/Robinson-Foulds.html > This will often be faster than dist.topo. > (As noted there, the RF distance has certain issues which means that > it is not always the most suitable measure of tree dissimilarity!) > And you can find ape's GitHub repo https://github.com/emmanuelparadis/ape > Martin > > ---------------------------------------------------------------------- > > Message: 1 > > Date: Mon, 06 Mar 2023 14:43:04 +0100 > > From: =?utf-8?B?Vm9qdMSbY2g=?= Zeisek <vo...@trapa.cz> > > To: mailinglist R <r-sig-phylo@r-project.org> > > Subject: [R-sig-phylo] Parallelization in ape::dist.topo > > Message-ID: <3551602.QO7bkq4lFn@veles> > > Content-Type: text/plain; charset="utf-8" > > > > Hello dear colleagues, > > I use often ape::dist.topo (see here dist.topo.r), which is doing > > the calculations sequentially, which is very slow for large data > > sets. I'm sorry, I haven't found any relevant Git repository or > > so, so I hope Emmanuel won't mind if I discuss it here. > > I discussed various options with ChatGPT and dist.topo.par1.r is the > > simplest solution, basically using mc.lapply instead of 2 for loops. > > Good study material for how to do it in general. Little enhancements > > are in dist.topo.par2.r, which should be slightly better in case > > some pair of comparisons would return NA or so, but from my tests > > there doesn't seem to be any difference. > > And finally there is dist.topo.par3.r which doesn't load parallel > > (and uses plain lapply) for cores==1, while parallel and doParallel > > for multiple cores. It also contains some checks and error handling. > > From my testing it works well. I'm not sure if tryCatch is really > > needed there. In any case, improvements welcomed. :-) > > So, what do You think? Is this usable improvement of ape::dist.topo? > > Sincerely, > > V. -- Vojtěch Zeisek https://trapa.cz/en/ Department of Botany, Faculty of Science Charles University, Prague, Czech Republic https://www.natur.cuni.cz/biology/botany/ https://lab-allience.natur.cuni.cz/ Institute of Botany, Czech Academy of Sciences Průhonice, Czech Republic https://www.ibot.cas.cz/en/ Computing cluster https://sorbus.ibot.cas.cz/en/start
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ 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/
