Running ML search on bootstrap file

[Mark Quinton-Tulloch]

I'm trying to work out how to run the ML search on the file of bootstrap trees.

I initially ran RAxML as follows:

raxmlHPC-PTHREADS-SSE3 -s gene_align.phy -n raxml.gene.out -m GTRCAT -T 40 -f a -x $RANDOM -N autoMRE -p $RANDOM -U

However, I reached the server timeout limit after the bootstrapping stage.  As such I have the RAxML_bootstrap.raxml.gene.out file, and would like to pick up the analysis from this point.  Is there any way of doing this?  My searches suggest that I should use the '-f d' option, but this seems to require the original alignment file as an input, not the file of bootstrap trees.

Any help would be greatly appreciated. 

[Alexandros Stamatakis]

Hi Mark,

First, we recommend using the all new RAxML-NG from now on:

https://github.com/amkozlov/raxml-ng

Second, is you MSA long enough to run efficiently with 40 threads?

There should be at least 1000 sites per thread, i.e., 40,000 sites for
this to run efficiently.

You can run the ML tree search with -f d what is read in in your command
line below is the original MSA anyway, the bootstrapping only happens
internally in the code.

Hope that helps,

Alexis

> --
> You received this message because you are subscribed to the Google
> Groups "raxml" group.
> To unsubscribe from this group and stop receiving emails from it, send
> an email to raxml+un...@googlegroups.com
> <mailto:raxml+un...@googlegroups.com>.
> For more options, visit https://groups.google.com/d/optout.

--
Alexandros (Alexis) Stamatakis

Research Group Leader, Heidelberg Institute for Theoretical Studies
Full Professor, Dept. of Informatics, Karlsruhe Institute of Technology

www.exelixis-lab.org

[Grimm]

Hi Mark,

As Alexis said, you can just infer the tree using -f d; which is an independent step at the end of the bootstrapping (note that the final tree optimisation will anyway re-optimise the model when you use the "CAT" models – the per-site rate approx. is only used for the bootstrapping step).

When you have your final tree with -f d (in the RAxML you use, you could also invoke the usual final step with -f T, I suppose, which does "final thorough optimization of ML tree from rapid bootstrap search in standalone mode"), you just use the following command to map the BS support from your finished bootstrapping analysis onto the optimised tree:

-f b -m GTRCAT -t RAxML_besttree... -z RAxML_bootstrap.raxml.gene.out -n xxx

Cheers, Guido

[Mark Quinton-Tulloch]

Thanks for your help.  I've switched to RAxML-NG for my analysis.

> an email to ra...@googlegroups.com
> <mailto:ra...@googlegroups.com>.