How to force bcftools to call all variants

Question

[Hans]

Hello I am using bcftools to call variants with this command:

bcftools mpileup   -Ou -b bamlist  -f ref.fasta  | bcftools call -Ob    -mv   >variant.bcf

However, for some specific variants that I know to exist (looking at bam files with IGV), I do not get variant calls. The read depth is shallow 3-4 reads per site. How do I force bcftools to output all possible variants, even tough they have low probability?

[stefania.morelli92]

Hi,

I have the same problem and i would also like to force call bcftools in order to get all the variants of my bam. also i am aware of some variants (examining the bam files with Tablet) but they are not called

I am using this command

bcftools mpileup -B -q30 -Q30 -f reference.fasta -a FORMAT/DP,FORMAT/AD --threads 6 -R list_of_specific_position.txt file.bam | bcftools call -m -f GQ -O v -o call_variant.vcf

I tried to apply the command you shared but it creates a file that I can't read. How can I do it? Do you have any suggestions?

tks for the help

Stefania

[GenoMax]

You should start relaxing the filters (if it was not apparent from the comment I added in other post you created). You are filtering things with -q and -Q options.

Answer 1

[Istvan Albert]

There are some default parameters set like p-value < 0.5 and potentially others. Try setting these to higher values.

bcftools call -h

Another alternative is to not put the -v there at all, that way you'll get the genotype likelihoods for all positions (potentially a huge file)

[Hans]

I was trying to use the p-value > 0.5 option which is available only under the -c option, but it did not solve the problem.

I have added the FORMAT/AD and DP to the output, omitted the -v flag and added bcftools filter to the pipe and got the desired result. I can further look at the AD (allele depth) to decide which SNP is true.

bcftools mpileup  -a FORMAT/AD,FORMAT/DP  -Ou -b bamlist  -f ref.fasta  | bcftools call -Ou    -m   | bcftools filter -Ob -i 'F_MISSING<0.1&&MAF>0.1'  >variant.bcf

[Istvan Albert]

thanks for sharing the solution.