Variant calling frequency using Samtools

Question

[ronen.levy]

I am trying to identify the polymorphism in my data using the following commnad:

samtools mpileup -uf <refernce_file> <sorted_bam_file> | bcftools-1.2/bcftools call -cv -Ov > output_file

At the next stage, I filter the output based on DP>10 && QUAL>30.

However, before the filtration stage I realized that some of the polymorphisms were not identified. What could be the reason?

Answer 1

[Manuel Landesfeind]

Samtools/Bcftools implement a statistical model to calculate the likelihoods for all genotypes of a variant given the data (i.e., observed sequencing reads). If this calculation results in 0/0 being the most likely genotype, the variant is considered "wild-type"/"non-existent" and will not be reported. You can overcome this by dropping the "-v" parameter or by setting "--pval-threshold" parameter to a lower probability. However, this will also introduce additional variants which you may consider as false positives...