Error with gatk 4.6.0.0 gvcf

[fsandron]

Hello,
I tried using glnexus_cli but it fail to parse some gvcfs from gatk 4.6.0.0 (haplotypeCaller and ReblockGVCF).
For example

Invalid: allele is not a DNA sequence  (reblocked_gvcfs/SAMPLE_hg38_gatk4.reblocked.g.vcf.gz <NON_REF> chr1:123533581-123533581)

The input gvcf entry look like this

#CHROM  POS     ID      REF     ALT     QUAL    FILTER  INFO    FORMAT  SAMPLE
chr1    123533581       .       T       <NON_REF>,C     31.23   .       AS_QUALapprox=|0|66;QUALapprox=66;RAW_GT_COUNT=0,1,0;RAW_MQandDP=0,0;VarDP=75   GT:DP:GQ:MIN_DP:PL      0/2:75:0:75:66,1299,1339,0,1748,40

I guess that the "<NON_REF>" before the "C" in the ALT field could trigger the issue (because it's usually the other way around)
This issue may be linked with #299

[Ons14]

Hello

I am facing the same problem, I am working with gvcf files generated with GATK 4.2.3.0 and I'm seeing the same error message at positions containing <NON_REF> :

chr1 123533571 . G <NON_REF>,T 0 . END=123533571 GT:DP:GQ:MIN_DP:PL 0/0:41:54:41:0,111,1386,48,1322,1316

chr1 123533619 . T <NON_REF>,C 0 . END=123533619 GT:DP:GQ:MIN_DP:PL 0/0:47:37:47:0,114,1142,58,1087,1081

Have you been able to find a solution to this? I'd really appreciate any guidance you can share.
Thanks for your time and help!

[fsandron]

Hello,

I ended up not using GLnexus and used gatk directly to make the global vcf.
However, GLnexus worked fine on my deepvariant output.

You could probably try reordering ALT to have <NON_REF> in last place but I didn't bother.

Sorry I can't provide a better solution.

[Ons14]

Thank you very much for your response!
I'm already using gatk, but I was looking for a faster tool, gatk is taking a very long time ...

[kvn95ss]

I noticed reblocked gvcf is used, maybe this could be causing the errors? How about trying a non-blocked gvcf and seeing if this helps?