Re: [wgs-assembler-users] cgw running single core?

[Brian W. <th...@gm...>]

That was one of the pieces we could never effectively thread, due to a
large quantity of thread-unsafe code.  There was some work starting in
CA8.0 to better prune the merges attempted, enabled with
cgwMergeFilterLevel=2 (or 5, but scaffolding suffers).

You can also increase the minimum number of mate pairs required for a
scaffold join with cgwMinMergeWeight=2 (the default).  This equates to cgw
opiton -minmergeweight.

I don't think masurca supports either though.

With manual intervention, you can stop scaffolding at any time and move on
to the next algorithmic step.  The process is hopefully described
adequately at:

http://wgs-assembler.sourceforge.net/wiki/index.php/Scaffolder_failure

under 'force it to recompute' about half way down the page.  The summary is
to kill the existing cgw, edit the 7-0-CGW/*timing file to change the
'(logical ckp*)' to '(logical ckp05-1SM)', and restart.  This forces cgw to
restart from the algorithmic step after the one it is stuck on.

b


On Mon, Apr 13, 2015 at 3:49 PM, mathog <ma...@ca...> wrote:

> Hi,
>
> I have been running a bunch of MaSuRCA jobs, testing how well it does
> with different simulated library combinations on simulated homozygous
> and heterozgyous (4% difference) genomes using only Illumina data.  The
> homozygous ones finish in about 8 hours. The one heterozygous one to
> complete to date took 29 hours. Almost all of the extra time was spent
> with cgw running single threaded, or at least it hovers at just around
> 100% CPU, which is pretty low on a 46 core machine.  So the issue seems
> to be in the (modified) wgs assembler, not in the steps that MaSuRCA
> runs before that.  An earlier experiment with real(ly bad) data saw cgw
> run for 29 _days_ before MaSuRCA crashed - almost all of that as cgw
> running on one core.
>
> Is there a way to speed cgw up?  Some command line switch which has been
> omitted or set incorrectly? Here is the cgw command on the diploid test
> that is running now (which is currently showing 6 hours of cumulative
> CPU time, and has 8.4G resident):
>
> /opt/MaSuRCA/CA/Linux-amd64/bin/cgw -j 1 -k 5 -r 5 -s 2 -z -P 2 -B
> 167006 -m 100 -g
> /home/mathog/wgs_project/do_masurca_diploid/CA/genome.gkpStore -t
> /home/mathog/wgs_project/do_masurca_diploid/CA/genome.tigStore -o
> /home/mathog/wgs_project/do_masurca_diploid/CA/7-0-CGW/genome
>
> The "diploid" genome in this experiment is C elegans, with one wild type
> copy and one mutated copy.
>
> Thanks,
>
> David Mathog
> ma...@ca...
> Manager, Sequence Analysis Facility, Biology Division, Caltech
>
>
> ------------------------------------------------------------------------------
> BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT
> Develop your own process in accordance with the BPMN 2 standard
> Learn Process modeling best practices with Bonita BPM through live
> exercises
> http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual-
> event?utm_
> source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF
> _______________________________________________
> wgs-assembler-users mailing list
> wgs...@li...
> https://lists.sourceforge.net/lists/listinfo/wgs-assembler-users
>