[mira_talk] Re: denovo assembly runs out of memory

On Freitag 12 Juni 2009 Andreas Petzold wrote:
> [...]
> To us it seems that the crash happens at the graph phase so maybe we have
> to much spurious overlaps. Could they somehow be reduced ? Should we
> adjust the parameters ?

Hi Andreas,

it's not really a graph phase there. This phase stores read pairs that don't 
belong together and this actually can be a problem for some genomes with lots 
and lots of repeats.

I'm afraid, you will have to mask, Can you please read through
  http://www.freelists.org/post/mira_talk/RfC-Getting-mean-genomes-assembled
which I just posted?

In MIRA versions >= 2.9.45 (I'd recommend *45x3) you will find a short report 
on "hash statistics" in the output. Could you please report it so that I have 
another example for a fungal genome? Thanks.

> Furthermore we have access to a 64bit Itanium cluster with 146 GB RAM (os:
> linux). Is there a possibility to provide a fitting binary so that we can
> use mira on this cluster ?

I'll come back to that shortly.

However, the last time I saw run times on an Itanium system I was ... a bit 
perplex. Some routines ran *way* slower than on AMD64 or Intel64 CPUs. I guess 
the Itaniums did not like the memory addressing scheme used in some MIRA 
algorithms.

Regards,
  Bastien


-- 
You have received this mail because you are subscribed to the mira_talk mailing 
list. For information on how to subscribe or unsubscribe, please visit 
http://www.chevreux.org/mira_mailinglists.html

Other related posts: