31_CoriellNA12878.gatk.vcf.gz upload failed

[davmlaw]

See:

http://144.6.229.160/upload/view_upload_pipeline/1149

I've also made a small VCF that reproduces this (need to unzip before you upload) https://app.zenhub.com/files/299486514/9b9563b2-8575-4c54-a58e-28b811f55acb/download

#CHROM	POS	ID	REF	ALT	QUAL	FILTER	INFO	FORMAT	31_CoriellNA12878
chr13	37564190	rs1253322554	A	.	109.74	PASS	AN=0;BaseQRankSum=0.674;DB;DP=129;ExcessHet=0.0025;FS=0;InbreedingCoeff=0.3955;MQ=60;MQRankSum=0;QD=27.07;ReadPosRankSum=-0.674;SOR=1.609;VQSLOD=8.09;culprit=MQRankSum	GT:AD:DP:GQ:PGT:PID:PL:PS	.|.:0:.:.:0|1:37564182_A_T:.:37564182

  File "/mnt/variantgrid/upload/vcf/vcf_import.py", line 229, in import_vcf_file
    bulk_inserter.process_entry(v)
  File "/mnt/variantgrid/upload/vcf/bulk_genotype_vcf_processor.py", line 231, in process_entry
    phred_likelihood_str = self.get_phred_likelihood_str(variant)
  File "/mnt/variantgrid/upload/vcf/bulk_genotype_vcf_processor.py", line 187, in get_phred_likelihood_str
    pl_value = int(pl[i][pl_index])

[davmlaw]

Reopening because it was auto-closed by git commit.

This bug popped up here for the first time as we expected PL to be a 3 element array, but it was "." as the sample had a genotype of ".|." which we had never seen before. Made special case code to handle that

Redeployed, reloading VCF on vg test... failed again with:

  File "/mnt/variantgrid/upload/vcf/bulk_genotype_vcf_processor.py", line 191, in get_phred_likelihood_str
    pl_index = pl_index_lookup[gt]

chr7	42924353	rs552132089;rs771560161;rs869187830;rs370736797	CAA	.	57.23	PASS	AN=0;DB;DP=95;ExcessHet=0.0045;FS=0;InbreedingCoeff=0.2568;MQ=60;POSITIVE_TRAIN_SITE;QD=28.61;SOR=2.833;VQSLOD=4.88;culprit=FS	GT:AD:DP:GQ:PGT:PID:PL:PS	.|.:0:1:.:0|1:42924353_CAAA_C:0:42924353

Note: This has an ALT of "." and GT=".|." - but cyVCF2 calls it a zygosity of HOM_ALT here

[davmlaw]

Raised CyVCF2 issue brentp/cyvcf2#187 - will write workaround

[davmlaw]

I think Frank has fixed this by removing samples with ".|." via commit:

https://bitbucket.org/sacgf/ngs-pipelines/commits/eeb43fbff5137121fe6b5ac824bdbe3083aa5d51

But need to confirm

[davmlaw]

Frank confirms pipeline change ensures we won't see this zygosity again - no need to deal with this, will re-open again if ever happens in another file