Skewed sex distribution of SVs in PAR1

Structural Variation
1

Hi,

We are interested in the DHRSX gene, located in the pseudoautosomal region of X/Y chromosomes and recently were contacted with a case of a heterozygous deletion encompassing Exon 2. There are two deletions of a similar size and location in SV v4.1: gnomAD ; gnomAD

These were called using various tools, but what caught our eye was: 1) the fact that there are homozygous deletions, unexpected in this very severe disorder; 2) there is a very strange sex distribution of these deletions. In one case 30/31 alleles are in XY individuals, essentially impossible by change, unless I am overlooking something biological. I wondered if there is a technical explanation for this e.g. miscalling in the PAR in certain cases?

Best wishes,

Matthew

2

Just bumping this: I would appreciate someone looking into it!

Thanks,

Matthew

3

Hi Matthew,

Apologies for the delayed reply. We revisited these variants and traced them back to the raw data. Based on the current evidence:

a. DEL_CHRX_A32E6267 appears to be a true deletion. One carrier can be manually confirmed with paired-end, split-read, and read-depth support. The remaining samples show only read-depth support, which is less reliable in PAR regions (see below).

b. DEL_CHRX_B990CE03 shows sequencing depth support across all carriers.

There are a couple of GATK-SV settings that may reduce the reliability of deletions and duplications in the 5–20 kb range in PAR regions:

  1. Our pipeline requires sequencing depth evidence to confirm deletions/duplications >5 kb, and variants are marked as “PASS” if depth support is present—even when it is the only supporting evidence. While sequencing depth is generally a robust metric, smaller CNVs (<20 kb) are more prone to false positives when supported by depth alone.

  2. Sequencing depth is inherently less reliable in PAR regions due to ambiguous alignment of short reads in these regions.

I hope that makes sense, but let me know if we can further support.