[Molecular-evolution] 16S rRNA based phylogeny and fastDNAmlp looping problem

Tõnu Margus tmargus at ebc.ee
Tue Jun 6 11:03:52 EST 2006


Dear all,

 

I try to build a 16S rRNA based  phylogenetic tree for 160 bacteria and use
for this purpose the parallel version of  fastDNAml  v1.2.2p (Olsen etal.
1994).  Serial version of fastDNAml  runs ok (it takes overnight), but if I
try to use parallel version  (mpi or pvm) then It can not exceed  limit of
130 sequences even during additional 48 h. 

 

The README warns that there is possibility to go to loop, when “worker”
timeout exceeds 120 sec. then “foreman” had  to decide that “worker” has to
gone and gives this job to some new “worker”. 

 

It did not seems to be a case. 

I run it with option  -d 4  and monitor shows how trees are coming and going
and amount of evaluated trees are exceed 3 000 000.

 

pvm_dnaml_mon     -d 4    -p 16   -n 16srRNA    -w /home/tmargus/SSU/  -s
infile

 

I have increased timeout  from 120 sec to 1200 sec  and recompiled parallel
version, but without any success.

 

Had someone face similar problem(s)?

Has someone good idea how to solve it?

 

 

 

Thanks in advance,

 

Tonu Margus

 

 

 

 

 

 

 

 

 

 

========================================

Tonu Margus

Department of Bioinformatics

Institute of Cell and Molecular Biology

University of Tartu

Riia str. 23, Tartu 51010, Estonia

tel: +372-7376037

email: tonu.margus at ebc.ee

web: http://bioinfo.ebc.ee/ 

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.bio.net/bionet/mm/mol-evol/attachments/20060606/0a6c73a9/attachment.html


More information about the Mol-evol mailing list