Evaluating RMSMappingQuality profile to set threshold

Question

[serpalma.v]

Hello!

I am trying to set a threshold to filter my VCF file by using, among other INFO-annotations, RMSMappingQuality (I am using the GATK best practices workflow).

According to GATK, a reasonable cutoff value for this annotation is >=50, however, given the RMSMappingQuality profile in my data, that leaves out too many SNPs with lower scores than 50(please follow link to figure).

I was wondering what could be a source of the considerable amount of SNPs having low RMSMappingQuality.

Thanks!