Re: [wgs-assembler-users] cgw error

[Walenz, B. <bw...@jc...>]

There have been infinite loops here in the past, but they are somewhat rare, and I think were before the 6.1 release.

If you want to intervene, restarting at the checkpoint after the merge (there are two merge steps) will work.  If this is the last iteration of cgw (7-4-CGW typically) then care must be taken to not skip too much, otherwise pieces of the algorithm are skipped.  I’d just let it run here.  The CGW in MaSuRCA was modified to be less aggressive at scaffolding, and is generally quick.

The error itself can cause rare (again) issues with gap sizes, but is otherwise harmless.  We’ve seen a few assemblies with enormous gap sizes, far far larger than the mates would support.  Nearly every assembly hits this problem.

b


On 6/20/13 11:41 AM, "Ben Elsworth" <el...@gm...> wrote:

Hi,

I'm running celera 6.1 inside MaSuRCA and I appear to have a never ending CGW condition which keeps giving this error:

WARNING: variance difference is negative -- probably trouble with variances after interleaving
WARNING: variance difference is negative -- probably trouble with variances after interleaving
WARNING: variance difference is negative -- probably trouble with variances after interleaving
WARNING: variance difference is negative -- probably trouble with variances after interleaving
SOMEBODY IS SCREWING UP SCAFFOLDING -- RecomputeOffsetsInScaffold has a singularity -- assert skipped!
ReomputeOffsetsInScaffold failed (1) for scaffold 898025 in MergeScaffolds
MergeScaffoldsAggressive()-- iter 312 -- continue because we merged 3 scaffolds. minWeightThreshold = 6.9 maxAllowedIterations = 16
* CleanupScaffolds through scaffold 90000
* CleanupScaffolds through scaffold 120000
* CleanupScaffolds through scaffold 150000
* CleanupScaffolds through scaffold 160000
* CleanupScaffolds through scaffold 170000

I've tried stopping it and starting at a later checkpoint but it still seems to be stuck. Will this eventually finish or should I intervene?

Cheers,

Ben