在 R（ape 库）中根植基于简约的树时出错

Question

I am performing bootstrap analysis in R using maximum parsimony algorithm (pratchet() in ape library). When I run the analysis on unrooted trees (produced using pratchet() function), the bootstrapping runs fine. But when I want to root each bootstrapped tree before finding bootstrap support, I get error in rooting randomely at any of 100 trees. Note that this occurs before calling any code for computing bipartitions or clade support.

If I use neighbor joining algorithm (nj() in ape), there is no problem in rooting at all or in downstream bootstrapping, but apparently it occurs (randomly) while rooting parsimony-based trees using an outgroup. The strange thing I observed is that if I write the unrooted trees to file before rooting them (in case error occurs while rooting) and later want to root them, it works perfectly fine.

Here is the code, I am using for analysis.

performBootstraping = function(charMatrix, bsIterations) {
    # charMatrix is a DNA alignment matrix
    # bsIteration are number of bootstrap iterations.

    library(phangorn)

    phySeq = phyDat(charMatrix)

    treeMPRooted = getRootedParsimonyTree(charMatrix)

    bValuesMP = boot.phylo(treeMPRooted, charMatrix, FUN=function(xx) {tt = getRootedParsimonyTree(xx); return(tt) }, 
    B = bsIterations, trees=T, rooted=T)

   # convert to percentage
   bValues = bValuesMP$BP/bsIterations * 100;

   plot(treeMPRooted, use.edge.length = F); 
   title('Max Parsimony tree with bootstrap percentage')
    nodelabels(bValues, frame = 'rect')

    # write the tree as newick 
    write.tree(treeMPRooted, paste0(outDir,'/rooted_MP.tree'))
    return(bValuesMP)
}

getRootedParsimonyTree = function(cMatrix) {
    phySeq = phyDat(cMatrix);
    treeMP = pratchet(phySeq)
    treeMPRooted = root(treeMP, outgroup='Germline', resolve.root=T)

    return(treeMPRooted)
}

Here is the stack trace and the error

 Error in phy$edge[sndcol, 2] <- newNb[phy$edge[sndcol, 2]] <- n + 2:phy$Nnode : 
  number of items to replace is not a multiple of replacement length 
6 root(treeMP, outgroup = "Germline", resolve.root = T) at libaryFunctions.R#105
5 getRootedParsimonyTree(xx) at libaryFunctions.R#32
4 FUN(x[, boot.samp]) 
3 boot.phylo(treeMPRooted, t.vaf, FUN = function(xx) {
    tt = getRootedParsimonyTree(xx)
    return(tt)
}, B = bstrapCount, trees = T, rooted = T) at libaryFunctions.R#32
2 performBootstraping(vaf, outDir, i, bsIterations) at runAllSitesBootstrapForAllPatients.R#15
1 runAllSitesBootstrapForAllPatients(ccfFileDir = ccfDir, outDirPref = outDir) 
In addition: Warning message:
In newNb[phy$edge[sndcol, 2]] <- n + 2:phy$Nnode :
  number of items to replace is not a multiple of replacement length

Answer 1

I got frustrated with the 'root' function so edited the ape source code to create a function that roots a tree without throwing errors. It's included in the package 'TreeTools' as functionRootTree() , which also implements a few other options for setting the root.

Install it to R using:

install.packages('TreeTools')
RootTree(tree, outgroup)

It doesn't support edge lengths, I'm afraid.

Answer 2

look at the examples of the function bootstrap.phyDat and plotBS in phangorn, should simplify your function. Furthermore you don't need to root the individual bootstrap trees, just the tree you want to draw. This should solve your problem.

Regards Klaus