[wgs-assembler-users] Make-consensus errors in pacBioToCA

[Ole K. T. <o.k...@ib...>]

Hi,
I'm correcting some XLXL PacBio reads, and hit upon some failed assertions
in make-consensus. I hope it's possible to solve.

This is the first error (last 10 lines of the out file from the
runPartition.sh):
  with 100 errors (99.01% error)
In contig 814 forced alignment of 65 bases of string 48036185 subscript 344
to 2 bases of consensus
  with 64 errors (98.46% error)
In contig 814 forced alignment of 398 bases of string 52752554 subscript
345 to 1 bases of consensus
  with 397 errors (99.75% error)
In contig 814 forced alignment of 227 bases of string 56213438 subscript
346 to 18 bases of consensus
  with 214 errors (94.27% error)
In contig 814 forced alignment of 112 bases of string 75693310 subscript
347 to 2 bases of consensus
  with 110 errors (98.21% error)
make-consensus: align.cc:4903: void Complete_Align(const char*, int, int,
const char*, int, int, int, int, int, int, int, Align_Score_Entry_t*,
Align_Score_Entry_t&, Alignment_t&): Assertion `t_lo < t_slip' failed.

98.21 % error is a bit weird. Got any idea what that assertion is?


This is not an error, but even weirder:
  with 62 errors (151.22% error)
In  Reset_From_Votes  in contig 1369
  Forced alignment of string subscript 685 to consensus
  with 169 errors (174.23% error)
In  Reset_From_Votes  in contig 1369
  Forced alignment of string subscript 686 to consensus
  with 154 errors (179.07% error)
In  Reset_From_Votes  in contig 1369
  Forced alignment of string subscript 687 to consensus
  with 78 errors (185.71% error)

185.71 % error? That's a lot.

This is another error:
In contig 1135 forced alignment of 82 bases of string 52574042 subscript
809 to 11 bases of consensus
  with 74 errors (90.24% error)
In contig 1135 forced alignment of 225 bases of string 41625223 subscript
810 to 3 bases of consensus
  with 222 errors (98.67% error)
In contig 1135 forced alignment of 146 bases of string 55029895 subscript
811 to 2 bases of consensus
  with 144 errors (98.63% error)
In contig 1135 forced alignment of 140 bases of string 40943538 subscript
812 to 3 bases of consensus
  with 137 errors (97.86% error)
ERROR:  Bad from = 3 in Global_Align
  in file  align.cc  at line  6872  errno = 0

How can I get this kind of errors in the alignments?

The genome is about 830 Mbp (it's cod). Correcting with 454 reads, about
27x total. This is about 3.5x coverage in PacBio reads:
Running with 2536048074 bp for pblr_XLXL_cod.
Correcting with 22252248087 bp.

Could something have gone wrong in the correctPacBio step?

Thank you.

Ole