Re: [wgs-assembler-users] cgw error

[Ben E. <el...@gm...>]

Hi Brian,

Thanks again for your help. I got a little further this time. Now I have an
error again during the cgw stage but this time the error
from 7-2-CGW/cgw.out is:

* FOEXS: SUSPICIOUS Overlap found! Looked for (1184641,594183,O)[12,1073]
found (594183,1184641,I) 20
WARNING:  InsertChunkOverlap()-- Chunk overlap already exists.
NEW 594183,1184641,I - min/max 17/23 0/0 erate 0.250000 flags 10000 overlap
20 hang 0,0 qual 0 offset 0,7276224
OLD 594183,1184641,I - min/max 12/1073 12/1073 erate 0.250000 flags 10001
overlap 20 hang 310,300 qual 0 offset 0,7276224
WARNING:  CreateChunkOverlapFromEdge()-- Chunk overlap already exists.
 Keeping old overlap.
NEW 594183,1184641,I - min/max 17/23 0/0 erate 0.250000 flags 10000 overlap
20 hang 0,0 qual 0 offset 0,7276224
OLD 594183,1184641,I - min/max 12/1073 12/1073 erate 0.250000 flags 10001
overlap 20 hang 310,300 qual 0 offset 0,7276224
* Switched right-left, orientation went from O to I
* CreateAContigInScaffold() failed.
ContigContainment failed.
cgw: LeastSquaresGaps_CGW.C:1419: RecomputeOffsetsStatus
RecomputeOffsetsInScaffold(ScaffoldGraphT*, CIScaffoldT*, int, int, int):
Assertion `0' failed.

Any ideas?

Cheers,

Ben

On 30 May 2013 12:40, Walenz, Brian <bw...@jc...> wrote:

>  Hi, Ben-
>
> Somehow, various flags on reads are inconsistent.  Unless you’ve got a LOT
> of time invested in this cgw run, I’d recommend deleting the 7* directories
> and starting scaffolding again.  Is this near the start of the cgw process?
>  Can you get a stack trace?  If not, what does the *timing file contain?
>  Were there any oddities in the scaffold run so far?
>
> It’s failing because it wants to compute the insert size of a specific
> mate pair, but the two reads are in different contigs.  We can get around
> the current failure by explicitly ignoring this pair.  Right before the
> assert that fails, add:
>
> if (frag->contigID != mate->contigID)
>   continue;
>
> I.e., Doctor, it hurts when I do this.  Then don’t do that!
>
> b
>
>
>
>
> On 5/28/13 10:06 AM, "Ben Elsworth" <el...@gm...> wrote:
>
> Hi Brian,
>
> That didn't seem to help. I tried editing an earlier section of code in a
> similar way by adding this:
>
> if (mateContig == NULL){
>     continue;
> }
>
> below this:
>
> mateContig = GetGraphNode( ScaffoldGraph->ContigGraph, mate->contigID);
>
> But that led to another error:
>
> cgw: GraphCGW_T.C:3232: void ComputeMatePairStatisticsRestricted(int,
> int32, char*): Assertion `frag->contigID == mate->contigID' failed.
>
> Any idea what's going on?
>
> Cheers,
>
> Ben
>
> On 28 May 2013 11:00, Walenz, Brian <bw...@jc...> wrote:
>
> Hi Ben-
>
> I think this is harmless, and can be patched around.
>
> > To patch it up, and maybe avoid the crash here, add the following at line
> > 3206 in AS_CGW/GraphCGW_T.c
> >
> > if (extremeContig == NULL)
> >   continue;
> >
> > Line 3206 is just after "extremeContig = GetGraphNode(...)" and before
> the
> > call to GetContigPositionInScaffold().
>
> Line numbers above are relative to the latest code base.  In 6.1, unless
> masurca fiddled with GraphCGW_T.c, you want line 3296, in between these two
> lines:
>
>           extremeContig = GetGraphNode( ScaffoldGraph->ContigGraph,
> scaff->info.Scaffold.BEndCI);
>           GetContigPositionInScaffold ( extremeContig, &contigLeftEnd,
> &contigRightEnd, &contigScaffoldOrientation);
>
> The one other assembly that failed here (it was recent, too) finished
> successfully after the patch.
>
> b
>
>
>
> On 5/27/13 7:43 AM, "Ben Elsworth" <el...@gm... <
> http://el...@gm...> > wrote:
>
> Hi,
>
> I'm running v6.1 within MaSuRCA and keep getting this error during the cgw
> step:
>
> cgw: GraphCGW_T.C:3275: void ComputeMatePairStatisticsRestricted(int,
> int32, char*): Assertion `(mateContig) != __null' failed.
>
> It occurs after a lot of warnings about negative variance. I've tried
> following the advice here -
> http://sourceforge.net/apps/mediawiki/wgs-assembler/index.php?title=Scaffolder_failurebut keep getting the error.
>
> Any ideas?
>
> Cheers,
>
> Ben
>
>
>
>