Hi, I am hoping for some advice on the GnomAD v4 SV entry: DEL_CHR10_FD5D2313.
This variant has a Pass on the filtering stage indicating it is a high quality call, and a quality score of 76. I have been unable to find out what this quality score means and note that some other SVs have much higher quality scores. Is there any information available to explain what this quality score means and how it can be used to inform on the quality of the variant call?
Looking at the Genotype Quality plot, almost all the variant carriers have genotype quality scores at the low end of the scale, whereas the majority of other individuals have scores at >100. Could this indicate that this call is an artefact due to quality issues? Is it possible to determine whether or not this is a genuine copy number variant in these individuals?
Thank you for assistance with this.
Hi,
Thank you for raising this question. The quality score for each SV in our dataset is calculated as the median of the genotype quality (GQ) scores across all variant carriers. For homozygous alternate genotypes, we assign the maximum GQ value for this calculation. Please note that the original GQ values ranged from 0–999, and the SV quality scores were derived from those values. We later rescaled individual GQs to 0–99 but did not retroactively adjust the SV quality scores, which is why some quality scores exceed 100.
For this specific example, the variant is a 6 kb deletion supported solely by sequencing-depth evidence. Our pipeline consider deletions or duplications that are 5Kb or above with solely depth evidence as valid, if there are clear sequencing depth evidences. Three-quarters of the non-reference carriers are from PCR+ libraries, which typically have noisier depth profiles and therefore lower GQ, leading to a reduced overall quality score. In principle, this event should have strong depth support in the PCR− samples, and I would still consider the variant valid; however, the false-positive rate among PCR+ samples may be higher than expected.