Has anyone else ever randomly ran into the problem of misencoded quality scores with the GATK? This happened once before and now just recently again on a particular sample. The sample comes from the same batch as several other exomes. They were put through the identical pipeline. One ends up giving me an error about a misencoded quality (65) but seems to work fine if I add the fixMisencodedQuals flag. This is when using the DepthOfCoverage walker after marking Duplicates with Picard.
Pipeline is Trimmomatic -> BWA-MEM -> Picard SamFormatConverter -> Picard AddOrreplaceReadGroups -> Picard ReorderSam -> Picard MarkDuplicates
Two samples from the same batch went through the same pipeline. Anyone ever see quality scores get randomly misencoded like that before?