bcftools to remove non-variants sites

Question

[shiy05]

Dear all,

I am used to using vcftools to remove non-variants sites. The command I used:

vcftools --vcf input.vcf --mac 1 --recode --out input.mac

However, I cannot find the same command in bcftools.

I tried

bcftools view -m2 -M2 -v snps input.vcf

and

bcftools view --min-ac=1 input.vcf

All these commands keep more variants then I need, for example, the snp with all alternative genotypes 1/1.

However, I am able to get correct results using below command:

bcftools view --min-ac=1 -Q 0.9999 input.vcf

Any better suggestions?

[Jingjingzhang]

Hi, I've noticed that there seems to be a discrepancy between bcftools and vcftools when it comes to filtering out non-variant sites. I have a site that includes multiple alleles, with AC=27, 0, 19... vcftools --mac 1 will remove this site, whereas bcftools -c 1 will not.

vcftools --gzvcf xxx.vcf.gz --mac 1 --recode --recode-INFO-all --stdout | bgzip -c >xxx.mac1.vcf.gz
bcftools view -c 1 xxx.vcf.gz -Oz -o xxx.mac1.retry.bcftools.vcf.gz

xxx.vcf.gz: 713639 sites
xxx.mac1.vcf.gz: 129047 sites
xxx.mac1.retry.bcftools.vcf.gz: 161391 sites

Only the result of using bcftools can get the right vcf files.

[GenoMax]

vcftools is not actively maintained (last update was in 2018) and can thus be considered deprecated. You should use bcftools instead.

Answer 1

[shiy05]

I find out my answer:

bcftools view -c 1 -c 1:nonmajor test.vcf
#or
bcftools view -i "MAC >=1" test.vcf  #make sure you have only biallelic sites
#or
bcftools filter -e "MAC == 0" test.vcf #make sure you have only biallelic sites

However, if your %INFO tags is wrong, bcftools will produce error results.