[mira_talk] Re: assembling Nextera reads/MDA reads
- From: Bastien Chevreux <bach@xxxxxxxxxxxx>
- To: mira_talk@xxxxxxxxxxxxx
- Date: Mon, 18 Nov 2013 23:35:26 +0100
On 18 Nov 2013, at 21:32 , Walter, Mathias <mathias@xxxxxxxxx> wrote:
> […]
> I have a very closely related "reference" strain and plotted the
> insert size distribution (see attached pdf).
I think I start to not like Nextera. You’re the fourth report reaching me that
MIRA (even rc4) performs badly. In almost all other cases I could trace the
majority of the problems back to a *very* uneven Nextera coverage. There’s one
case pending as the bug itself is pretty nasty and needs further verification,
but for the others I am sure.
Would you have a plot of the coverage against the reference genome? Or better
yet, a wiggle file? I’m pretty curious what it looks like.
MIRA has never seen single cell / MDA data so far (at least not in my hands)
and if the data there looks like the one from the few articles I’ve seen,
things might look pretty grim. One idea (and it’s really just an idea) might be
to try in EST mode while switching off digital normalisation. But it will
introduce misassemblies for sure.
Last thing: having been bombarded with nasty projects where MIRA obviously
failed in the past few weeks, I sat down and fixed a lot of issues in the
assembly engine for some rare (and less rare) cases where the new 4.0 engine
failed.
Expect these improvements to be rolled out sometime this week (but Nextera will
remain a problem)
B.
Other related posts:
