New DRAGEN callsets alter the ploidy of samples by design - where imputed XY-individuals have their calls on sex chromosomes represented as haploid. That interrupts our prior methods for doing sex and relatedness checks.
The new approach (to be implemented) will be to check sex information in Seqr (as uploaded by PMs and collaborators, I believe) to the visible ploidy in the callset (which is easy enough to check, just seeing for each sample what percent of their reads are haploid - if it's anything greater than zero, then the sample was imputed as XY for biological sex) . Then, if anything here is discrepant between the two, then the PMs can refer to the sex & coverage docs that GP/DSP will deliver to determine which is correct.
matren395
commented
1 week ago
New DRAGEN callsets alter the ploidy of samples by design - where imputed XY-individuals have their calls on sex chromosomes represented as haploid. That interrupts our prior methods for doing sex and relatedness checks.
The new approach (to be implemented) will be to check sex information in Seqr (as uploaded by PMs and collaborators, I believe) to the visible ploidy in the callset (which is easy enough to check, just seeing for each sample what percent of their reads are haploid - if it's anything greater than zero, then the sample was imputed as XY for biological sex) . Then, if anything here is discrepant between the two, then the PMs can refer to the sex & coverage docs that GP/DSP will deliver to determine which is correct.
Closely tied to: https://app.zenhub.com/workspaces/tgg-methods-6613e5b68c36e00025172757/issues/gh/broadinstitute/tgg_methods/83