Variant annotations can be produced by HaplotypeCaller, Mutect2, VariantAnnotator and GenotypeGVCFs. The available annotations are listed in the Tool Index.
Note that some annotations values calculated by the different tools may be different for the same original data. There are a few things that generally account for differences in annotation values, linked to the normal behaviors of the tools. In all cases, you can end up looking at different sets or numbers of reads, which causes some of the annotation values to be different. It's usually not a cause for alarm. Remember that many of these annotations should be interpreted relatively, not absolutely.
HaplotypeCaller and Mutect2 apply a read realignment step that can modify local coverage counts and other annotation values that take coverage into account. In contrast, VariantAnnotator will calculate annotations either based on the pileup if you give it the original BAM file, or it will calculate summary metrics based on existing VCF fields. GenotypeGVCFs only calculates annotations based on existing VCF fields (that's pretty much its raison d'être).
Some tools apply different read filters by default.
Some tools apply downsampling in order to ensure good performance, but they may do so to different depths of coverage by default compared to others.
If there is anyway to "turn off the local realignment" in Mutect2 variant calling? Thanks.
Please sign in to leave a comment.