Fasta file not matching VCF

Question

[c.o.guldvog]

Hi,

I've used bcftools consensus to create fasta files from individual vcf files. However, some sites in the fasta file do not correspond to the vcf. Using the --sample flag resolved some of these problems, but not all. For example this site

chr22   2600340 .   C   T   1638.22 PASS    .   GT:AD:DP:GQ:PL  1/1:0,9:9:27:405,27,0

should have T, but

samtools faidx 8L19766_sam_sample.fasta chr22:2600340-2600340

yield a C.

The vcf files have been called using GATKs HaplotypeCaller. I have also attempted using GATKs FastaAlternateReferenceMaker to create fasta files, but this made the discrepancies far worse.

What could be causing these issues?

Thanks so much for any help!

Answer 1

[JC]

Your VCF it indicates reference is "C", and that's what you see in your Fasta.

To visualize the alternative call, you need to view your BAM file: samtools tview FILE.bam chr22:2600340-2600340

[c.o.guldvog]

No, if the VCF indicates that the individual has the alternative allele at a certain site, as is the case here, bcftools consensus should place the alternative allele at that position in the fasta file. This is how it works for most of my sites, but not all.

[finswimmer]

Do you have insertions or deletions in your vcf? If so, then chr22:260034 in the consensus fasta will not represent the same position in the original fasta.

[Ram]

Can you please give us the exact bcftools consensus command you're using?

[JC]

so, you create a consensus "patching" your reference genome? how do you generate that?

[c.o.guldvog]

No, there are no insertions or deletions .

This is the command I use:

cat house_sparrow_genome_assembly-18-11-14_masked.fa \
| bcftools consensus --sample 8L64869 8L64869_SNPs.vcf.gz \
> 8L64869.fasta