More posts like this
-
target vs. bait BED files for Illumina WES
written by Juan Cordero
-
no read counts per allele while using gatk ASEReadCounter
written by Sara
-
huge difference when comparing 2 vcf files from the same sample
written by Sara
-
low alignment rate of methyl-seq data
written by Sara
-
a lot of missing variants when using DRAGEN for variant calling
written by Sara
-
VCF evaluation using RTG.jar vcfeval
written by Sara
-
Problem with annovar in GATK4 output
written by ahmad mousavi
-
tools to get the coverage
written by Sara
-
CNVkit GRCh38 tumor-normal pair
written by plink_9857
-
Sciclone on exome sequencing data
written by cbst
Start by looking at the versions of GATK. There are substantial differences between newer GATK v.4.1.x and older versions. Are you using DRAGEN's implementation of GATK to do the end-to-end analysis? Did you use the latest GATK for manual analysis?
I think (if I recall correctly) Dragen started with GATK as a variant caller, but re-wrote it in the end to improve the performance. So maybe the key of the difference is that different algorithms were used.
https://gatk.broadinstitute.org/hc/en-us/articles/360039984151-DRAGEN-GATK-Update-Let-s-get-more-specific