I would like to obtain many "sub-samples" of a Maximum Likelihood tree by randomly removing tips (about 20-30% of them from the complete tree). However, unlike the drop.tips function that comes with ape, I was hoping that the branch lengths and branching order of the resulting tree would be re-adjusted so that it too has maximized likelihood. This would be the same result as a maximum likelihood tree constructed from a smaller subset of sequences - although ideally faster, as it must be done many (~100). Are there any tree-building tools that accomplish this? Perhaps through some efficient use of information obtained during the construction of the complete tree?
*I'm not super intrested in Bootstrap support values (although if there's a reasonably fast way to get those too, I'd be impressed and happy to use it)
*I'm aiming to work with sets of sequences numbering in the low to mid 1000s (~1.3 Kb long)
thanks for any help that can be provided!