RAxML: confusion about Bootstrap searches and ML searches

Question

[Picasa]

Hi,

Based on that documentation :

http://sco.h-its.org/exelixis/web/software/raxml/hands_on.html

And in particular, this command:

The nice thing about rapid bootstrapping is that it allows you to do a complete analysis (ML search + Bootstrapping) in one single step by typing

raxmlHPC -f a -m GTRGAMMA -p 12345 -x 12345 -# 100 -s dna.phy -n T20

If called like this RAxML will do 100 rapid Bootstrap searches, 20 ML searches and return the best ML tree with support values to you via one single program call.

1) I always thought that Bootstrap searches and ML searches was the same. So what is the difference ?

2) -# 100 means 100 bootstrap but from where the 20 ML searches come from ?

3) Usually should we not set the same value for these 2 ?

Thanks for your help

Answer 1

[ArtemP]

Hi!

AFAIK, ML search and bootstrap search are different things in RAxML. The ML search is performed to find the best-scoring ML tree among 20 ML trees calculated from different stepwise-addition parsimony starting trees.This is based on the original alignment. On the other hand, bootstrapping resamples the alignment positions randomly with replacement to the length of the original alignment (100 times in this case) and generates a single tree on each round to infer statistical support of the branches. Afterwards, the bootstrap values are placed on the corresponding branches of the best-scoring ML tree.

Does that make sense?

[hwc13642185115]

Hi! A relevant question from me: I'm clear about that bootstrap is different from ML search, but finally the "best-scoring ML tree" we get, is it the ML tree result from the original alignment, or the "best scoring" ML tree among all the trees, including the bootstrapped trees? Thanks!