[mira_talk] Re: PacBio for scaffolding?
- From: Robin Kramer <kodream@xxxxxxxxx>
- To: mira_talk@xxxxxxxxxxxxx
- Date: Mon, 18 Jul 2011 15:25:00 -0600
I doubt that pacbio is going to bring much to the assembly world until that
accuracy improves. The problem is the reads will still need to be "error
corrected" by competing sequencing technologies(similar to the problem that
454 reads have), and lipe data can handle up to 5KB insert sizes(sipe is at
800), furthermore the amount of data that is in a pacbio run makes it
questionable as to weather it is worthwhile to analyze for the sake of
assembly. The strengths of pacbio, absence of chemistry bias and strobed
reads, are something we are all looking forward to.
Sincerely yours,
Robin
On Mon, Jul 18, 2011 at 2:54 PM, Bastien Chevreux <bach@xxxxxxxxxxxx> wrote:
> On Jul 16, 2011, at 18:11 , 000.calabi.yau.000@xxxxxxxxxxxxxx wrote:
> > I have seen that PacBio released some E.coli datasets. (
> http://www.pacbiodevnet.com/share/datasets/EColiOutbreak).
>
> Yep, seen them too.
>
> > I wonder what your opinion is on using reads of this length for
> scaffolding in larger genome projects.
>
> Reads of length 3kb? Wonderful.
>
> > I mean the error rate seems pretty high still, but with such long reads
> this shouldn't be a too big problem, or?
>
> Unfortunately, it is. 15% error rate means an error every 6 to 7 bases on
> average. That's way too much to my likings. The normal MIRA workflow would
> also not work well, but I had plans to test a couple of things.
>
> > So I am wondering if one would think into that direction would it make
> sense to do a MIRA hybrid assembly or would this need more specialized
> assembly routines?
> > And if yes are thinking about adding support like this to MIRA?
>
> I am. Probably PacBio also realised that they would not get much momentum
> if many of the available tools do not work with their data. At least that is
> my interpretation of their recent efforts to present long CCS-reads
> (circular consensus sequence reads) which they say have 93% accuracy. Now,
> there's something MIRA can start to work with. Not really perfect, but
> anyway not bad.
>
> Will need some time though.
>
> 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:
