Merge VCFs with overlapping samples

Question

[Wan Shi Tong]

I have VCFs that have some overlapping samples, is there a tool that can do this...

###VCF1:
CHR POS ID ALT REF QUAL INFO FILTER FORMAT Sample1 Sample2 Sample3  
SNP1...  
SNP2...  
SNP3...  

---

###VCF2:
CHR POS ID ALT REF QUAL INFO FILTER FORMAT Sample2 Sample3 Sample4  
SNP2...  
SNP3...  
SNP4...  

---

I want this:

###VCF1+VC2:
CHR POS ID ALT REF QUAL INFO FILTER FORMAT Sample1 Sample2 Sample3 Sample4  
SNP1... (missing for Sample4)  
SNP2...  
SNP3...  
SNP4... (missing for Sample1)  

---

not this:

###VCF1+VCF2:
CHR POS ID ALT REF QUAL INFO FILTER FORMAT Sample1 Sample2 Sample3 Sample2_2 Sample3_2 Sample 4  
SNP1... (missing for Sample2_2, Sample3_2, and Sample4)  
SNP2...  
SNP3...  
SNP4... (missing for Sample1, Sample2, and Sample3)  

In this example of what I do not want, Sample2 and Sample3 would only have SNP1, SNP2, and SNP3 and Sample2_2 and Sample3_2 would have SNP2, SNP3, SNP4.

---

Is there a tool that can merge VCFs and keep only one copy of each sample?

[Kevin Blighe]

On face value, all that you require is bcftools merge. Pay close attention to the -m parameter, too. Missing genotypes will be represented as ./.

[zx8754]

merge would want to have unique samples over vcfs, we could use --force-samples but then we get suffixes which OP doesn't want.

[Wan Shi Tong]

Yeah, that is exactly my problem. vcf-merge and bcftools merge do not merge same samples. They create new entries for each repeated sample unfortunately.

[Kevin Blighe]

Would be easier to split these back into individual VCFs and then run bcftools concat --allow-overlaps --remove-duplicates to concat the same samples into a single VCF, and then merge everything with bcftools merge. This will work, as I have done it before for this type of situation.

Answer 1

[Getting there]

You need bcftools concat, I used the command below and got the result you described.

bcftools concat -a filtered_indels_annotated.vcf.gz filtered_snps_annotated.vcf.gz -Ov -o filtered_BC_merged.vcf

Some useful info here on the -a option: https://samtools.github.io/bcftools/bcftools.html#norm

[zx8754]

For concat to work we need all samples to overlap exactly.

All source files must have the same sample columns appearing in the same order.