genotype wrong

[lihaicheng7003]

./graphtyper genotype ~/data1/References/human/bowen/Homo_sapiens_assembly38.fasta --sam=NA12878/alignment/NA12878.alt_bwamem_GRCh38DH.20150718.CEU.low_coverage.bam --region=chr21 --threads=40  --prior_vcf=~/data1/References/human/dbsnp/dbsnp138/Homo_sapiens_assembly38.dbsnp138.vcf.gz  

I encountered an error, and I don't know how to solve it, constructor.cpp:1248 Found an SV in a non-SV graph at chr21:4395500

[lihaicheng7003]

I extracted chr21 from ~/data1/References/human/dbsnp/dbsnp138/Homo_sapiens_assembly38.dbsnp138.vcf.gz

bcftools view Homo_sapiens_assembly38.dbsnp138.vcf.gz chr chr21 > Homo_sapiens_assembly38.dbsnp138.chr21.vcf

Re-run, successful, and the time is much faster than the entire dbsnp

./graphtyper genotype ~/data1/References/human/bowen/Homo_sapiens_assembly38.fasta --sam=NA12878/alignment/NA12878.alt_bwamem_GRCh38DH.20150718.CEU.low_coverage.bam --region=chr21 --threads=40  --prior_vcf=~/data1/References/human/bowen/GATKResourceBundle/Homo_sapiens_assembly38.dbsnp138.chr21.vcf  --output=~/data1/project19/results_dpsnp138_chr21

[lihaicheng7003]

So I'm a little confused. If dbsnp is introduced into the entire genome, will mutations in other chromosomes affect chr21?

[hannespetur]

Hey,

Let me first say that --prior_vcf is typically not used. If it has too many nearby variants it will create a lot of graph paths which can make the mappings very ambigous and make the variant calling worse.

As for the error, please check how the VCF records look like at chr21:4395500. It could be malformed in some way.

For the chr21 rerun, you should add -Oz to the command to recompress the VCF and then tabix index it. I suspect reading the VCF is failing and graphtyper is genotyping as if it did not get any prior variants.

Best, Hannes

[lihaicheng7003]

Thanks for the reply, I will check again and reply later.