[mira_talk] Re: extremely high illumina coverage

  • From: "Bastien Chevreux" <bach@xxxxxxxxxxxx>
  • To: mira_talk@xxxxxxxxxxxxx
  • Date: Thu, 20 Oct 2011 18:42:09 +0200 (MEST)

Von: Ganga Jeena
> If someone can tell me how to do away with this GC bias in reads
> or a solution for it.

Careful: GC bias is something very different than the GGCxG problem.

Regarding the GGCxG problem, the best way to control it is to use in MIRA the 
proposed end clipping (-CL:pec), though I admit I need to refine it for data 
>>100x coverage.

Regarding a GC bias: that problem should be gone for good since beginning of 
2011 at the latest. If you have data from second half of 2009 till 4th quarter 
2010, you might have a GC bias and getting rid of that is a pretty daunting 
task, because it's almost impossible to tell the difference to regular repeats 
with simple kmer counting algorithms. One would need graph analysis and a lot 
of other tricks I think.

B.

Other related posts: