[mira_talk] Re: Lots of contigs, then segmentation fault
- From: Bastien Chevreux <bach@xxxxxxxxxxxx>
- To: mira_talk@xxxxxxxxxxxxx
- Date: Tue, 19 Apr 2011 16:19:49 +0200
On Apr 18, 2011, at 23:07 , Egon Ozer wrote:
> I must be doing something wrong...
Or not ...
> I'm trying to do a hybrid assembly project using 2,008,245 paired 454 reads
> and 5,722,894 PE Illumina reads. The Illumina reads are 53 bp (short, I
> know, but that's what I have to work with) and corresponds to about 45x
> coverage of my organism.
Together with longer reads like from 454 or Sanger, the length is less
important than in de-novo assemblies of Solexa only. I have a testbed at home
with 800k 454 FLX and 4m Solexa 36bp, and adding those Solexas performs true
miracles to the assembly.
> The program runs 6 and a half hours, then seg faults. When I check the log
> output, I see that it was working on contig # 24994 (!) when it died.
OK, so two problems. First one: MIRA should not segfault. Never. Ever. At. All.
Period.
> The seg fault is one problem which comes and goes. I have been able to run
> Mira to completion before on version 3.2.1.11 (after just restarting the run
> after the segmentation fault), but the bigger problem is that with that run I
> produced 80126 contigs, the largest of which was 1526 bases with an N50 of
> 97.
That is actually problem number two: something is not right with the data I
suppose. If I had to guess: the preprocessing of the SFF files went awfully
wrong. Did you use the right linker sequences?
Or some non-standard adaptor remained unclipped.
Could you send me the, say, first 2000 lines of the "log_assembly.txt" file?
I'd like to check a couple of things.
> Please give me a hand. I have been really counting on hybrid assembly to
> help me get as complete and accurate a sequence for my bacterium as possible,
> but I seem to be missing the mark on something.
If everything else fails ... do you think you could make the data available to
me so that I can have a more detailed look?
B.
--
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:
