Information about "How Many Bootstrap Replicates Are Necessary?"

[jaguilar]

Dear Dr. Stamatakis,

I am doing a phylogenetic analysis and I've read some about tree
support measures such as bootstrap, jackknifing, bremer support and
others. You published a paper "How Many Bootstrap Replicates Are
Necessary?" and there you propose some stopping criteria.
I am using the raxmlGUI 0.93 where in the reps options has autoMR,
autoMRE, autoMRE_IGN and autoFC.

I've searched at your website and on the RaxML Manual for information,
differences and comparison about this options but I couldn't get any
detail about what they do, in which situation I should one instead of
the others or any "standard" or suggested method for bootstrap
stopping.

I would be very very grateful if you could give me information about
them and advice me on how to use them for phylogenetic analysis. I am
using COI (800 bases) and ITS-1 (700 bases) for 97 taxons.

Thank you very much for your help.

I am looking forward to hearing from you soon.

Yours faithfully,

Juan Pablo Aguilar



P.S: Have you ever thought on including jackknifing for tree support
in Raxml?

[Alexis]

Hola Juan Pablo,

Regarding jackknifing: I will not implement this in RAxML since the
jackknife has no clear
statistical interpretation (see Felsenstein's book).

Regarding bootstopping:

If I remember correctly, we discuss the different approaches somewhere
in the paper.
In general, I wouldn't use FC, ideally you should use (according to
our experiments autoMRE) which is the most elaborate
method.
Here, the convergence check is based on extended majority rule
consensus trees which are however
a tiny bit hard/expensive to compute.
Thus, if you have datasets with many taxa (you can run your dataset
with 97 taxa with autoMRE)
of more than 1000 or 2000 taxa you should use autoMR or autoMRE_IGN
which is a dirty approximation of autoMRE that nonetheless seems
to work quite well.

Hope his helps,

Alexis