Question about BQSR
I am trying to figure something out in GATK, namely, BQSR. I am working an organism for which there are no known SNPs, so I ran the haplotype caller on uncalibrated data (realigned_reads.bam and raw_variants.vcf). Then I ran BQSR using the raw_variants.vcf file as the knownSites. According to my understanding, since the BQSR would SKIP re-calibrating SNPs from the knownSites file, the qualities for this variant in the before file (realigned_reads.bam) should be the same after re-calibration (recal_reads.bam). But as you can see from the color of the nucleotides, they are indeed lower QS (much lower) after recalibration. This is the source of my confusion. According to the docs, this site should have been skipped. Any clarification would be greatly appreciated.