Hi, Jaemin,
It seems like something is wrong - human chr1 (243Mb) runs for me in
about 3 hours, mostly cpu bound.
Human chr2 actually takes slightly longer than chr1, so it is possible
that there is some run time dependency on repetitiveness.
Is horse chr31 very repetitive? If so, then another thing to check for
is memory usage (i.e. is your process swapping excessively).
The java bindings for bwa will use a big chunk of non-heap virtual
memory in addition to the java heap.
You can run with a fairly small java heap (I think I use 1Gb), but it
will take additional memory approximately the size of the reference genome.
To answer your question directly, there is an unadvertised -L argument
that takes an interval (e.g. chr1:100-200) that would do what I think
you are asking for.
You would probably have to concatenate the resulting fasta files
together yourself, however.
But I would look first to see why your performance seems so slow.
-Bob
On 3/29/12 4:12 AM, Jaemin Kim wrote:
> Hi,
>
> I'm Jaemin Kim, a graduate student in Seoul National University in Korea.
>
> I'm trying to extract the CNV of horse genomes and I started with
> running Mask Files.
> I guess the computation is very heavy; it takes about 2 days to
> compute 25Mb chromosome 31.
> I see that it is allowed to compute chromosome by chromosome and then
> merge later.
>
> What if I even partition the chromosome region (for instance, by 10Mb)
> and then merge later? I know this would change the result of output
> Mask files, but do you think this would be significantly different?
>
> I need to make the computation as fast as possible, and so I need your
> advice.
>
>
> Thank you so much for your consideration!
> I will be waiting for your reply.
>
>
> Sincerely,
>
> Jaemin Kim
>
>
> ------------------------------------------------------------------------------
> This SF email is sponsosred by:
> Try Windows Azure free for 90 days Click Here
> http://p.sf.net/sfu/sfd2d-msazure
>
>
> _______________________________________________
> svtoolkit-help mailing list
> svt...@li...
> https://lists.sourceforge.net/lists/listinfo/svtoolkit-help
|
