[mira_talk] Re: 90% Roche454 reads going to debris file...help please !!

  • From: jyotsna guleria <jyotsna.guleria@xxxxxxxxx>
  • To: mira_talk@xxxxxxxxxxxxx
  • Date: Thu, 6 Oct 2011 17:11:22 -0400

Thanks so much !!

So, you suggest I should not use -SK:nrr=10 for 4k reads?
Can you suggest a range to use this option? e.g can I use it if I have 10k
reads or more?

For your question: any special reason you switch of proposed end clipping?
(-CL:pec=no)

We had a discussion about it long back where I mentioned that for some
reason if I turn ON this option of clipping I get more debris...don't know
why but it did happen and from then I have been assembling reads with same
command and never had any trouble so far...!



Many thanks for the help!

Jyotsna

On Thu, Oct 6, 2011 at 4:58 PM, Bastien Chevreux <bach@xxxxxxxxxxxx> wrote:

> **
>
> On Thursday 06 October 2011 18:07:28 jyotsna guleria wrote:
>
> > Can someone please help me in understanding what is going wrong by
> looking
>
> > at the logfile uploaded at :
>
> > All of my reads are going into debris list, I am not able to understand
>
> > why....
>
>
>
> 1. wrong command line. Do not write
>
> mira --project=... --job=... [some options] --highlyrepetitive ...
>
> but
>
> mira --project=... --job=... --highlyrepetitive [other options] ...
>
>
>  "--highlyrepetitive" is a position dependend parameter, look that up in
> the documentation.
>
>
>
> 2. any special reason you switch of proposed end clipping? (-CL:pec=no)
>
>
> 3. -SK:nrr=10 might be one cause for your problem. Almost every read has
> some
>
> kind of probelamtic stretch, see the log line:
>
>
>  "Writing read repeat info to:
>
> JCV345MDA_assembly/JCV345MDA_d_info/JCV345MDA_info_readrepeats.lst ...
>
> 3289 sequences with 8201 masked stretches."
>
>  Such a tiny project with 4k reads does not need that.
>
> 4. Only 1/3 of the reads is finally used in the assembly. See log lines
>
>  Building new contig 1
>
>      Localtime: Thu Oct  6 16:23:12 2011
>
>      Unused reads: 1514
>
>  The rest did not have any kind of overlaps. See again -SK:nrr
>
>  B.
>

Other related posts: