[mira_talk] Re: How does MIRA use paired end information?
- From: Evan <evan@xxxxxxxxxxx>
- To: mira_talk@xxxxxxxxxxxxx
- Date: Thu, 15 Sep 2011 14:54:16 -0700
So if MIRA has further passes to preform before it completes will it scan
through each contig and check for read pairs that are:
- misoriented
- bad distance
- online a single read made it into the contig (with the exception of
reads near the end of a contig oriented in such a way that its partner would
be past the end of the assembled contig)
And correct any inconsistencies by removing these reads from the contig and
placing them back into the pile of unassembled data?
On Thu, Sep 15, 2011 at 2:46 PM, Bastien Chevreux <bach@xxxxxxxxxxxx> wrote:
> On Sep 15, 2011, at 23:18 , Evan wrote:
> > Bastien, regarding step 2 if MIRA is extending a contig with a read who's
> partner is already contained within the contig being extended are both reads
> equally considered for removal? (IE, perhaps the new read isn't wrong but
> the first read was slightly misplaced)
>
> No, it stays. This can be considered both a strength and a weakness of
> MIRA.
>
> Strength: there are many libraries which are not, well, ideal in terms of
> template sizes. That way one looses only half of the reads. Furthermore, the
> read which was placed first was in 99.39845617% of all cases (a number I
> just made up now to say "almost always") a read which is not part of a
> repeat spanning more than the length of the read. That is: there should be
> only this one place in the genome where it could be placed.
>
> Weakness: in the remaining 0.60154382% of the cases (a number I just made
> up to say "sometimes shit happens"), the above proves to be wrong. Is likely
> to occur only at repeats with two copies in the genome and where sequencing
> coverage was (by misfortune) definitively more pointing to non-repetitive
> instead of repetitive.
>
>
> But chances are good that if the first read was misplaced in one of the
> passes of MIRA, that read (or a close one nearby) will trigger repeat marker
> base detection in case of misalignments, which will help MIRA in the next
> pass to avoid that misplacement.
>
> 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:
