phylotastic / datelife

R package containing datelife's core functionality
http://phylotastic.org/datelife/
13 stars 11 forks source link

what happens when we have no overlapping taxa in multiple chronograms? #51

Open LunaSare opened 6 years ago

LunaSare commented 6 years ago

We have two case scenarios:

LunaSare commented 6 years ago

For case 1, we added the find_groves functions. These look for sets of chronograms with at least two overlapping taxa. We have two cases here: a. there is just one grove b. there are several groves. Users can choose if they want the grove with the most taxa or the most trees. In both cases, all other chronograms are excluded from the summary

Sometimes, the best grove is complicated and summary methods do not work.

utils::data(names_subset2)
datelife_search(names_subset2, summary_format="phylo_sdm")
Error in solve.default(Q, col, 1e-19) : 
  system is computationally singular: reciprocal condition number = 3.23424e-24

Changing the tol limit will give a matrix from SDM, but will not work with hclust either, and will give the same error as using summary_format = phylo_median

datelife_search(names_subset2, summary_format="phylo_median")
Error in stats::hclust(DD, method = "average") : 
  NA/NaN/Inf in foreign function call (arg 11)