Open rrnewton opened 7 years ago
I have been running into this same issue and I was wondering if there have been any updates to resolve this? Thanks!
` Cleaning away previous phybin outputs... Parsing 24200 Newick tree files.
Total unique taxa (5): tx1 tx2 tx3 tx4 tx5 Note: defaulting to expecting ALL 5 to be present..
Number of input tree files: 24200 Number of VALID trees (correct # of leaves/taxa): 24200 Total tree nodes contained in valid trees: 121000 Average branch len over valid trees: 0.0 Max/Min branch lengths: (0.0,0.0) Creating equivalence classes (bins)... phybin: Internal invariant broken. These two were equally good roots: NTInterior (StandardDecor 0.0 Nothing) [NTLeaf (StandardDecor 0.0 Nothing) 4, NTInterior (StandardDecor 0.0 Nothing) [NTLeaf (StandardDecor 0.0 Nothing) 3, NTInterior (StandardDecor 0.0 Nothing) [NTLeaf (StandardDecor 0.0 Nothing) 2, NTInterior (StandardDecor 0.0 Nothing) [NTLeaf (StandardDecor 0.0 Nothing) 0, NTLeaf (StandardDecor 0.0 Nothing) 1]]]] NTInterior (StandardDecor 0.0 Nothing) [NTLeaf (StandardDecor 0.0 Nothing) 4, NTInterior (StandardDecor 0.0 Nothing) [NTInterior (StandardDecor 0.0 Nothing) [NTLeaf (StandardDecor 0.0 Nothing) 3, NTInterior (StandardDecor 0.0 Nothing) [NTLeaf (StandardDecor 0.0 Nothing) 2, NTInterior (StandardDecor 0.0 Nothing) [NTLeaf (StandardDecor 0.0 Nothing) 0, NTLeaf (StandardDecor 0.0 Nothing) 1]]]]] `
(Reported by Stefanie Hartmann)
This has to do with an internal algorithm that Phybin uses to pick "canonical" roots so that comparing unrooted trees becomes as easy as comparing rooted trees.
Clearly the algorithm has failed in this case: