Why is WGS/WES depth of Y chromosome highly variable?

Question

[samuelandjw]

Edit 1: The following is Y/Auto depth versus age in a cohort. Individuals in this cohort are not known to be cancer patients.

Original question:

In WGS/WES studies I read and studies I participated in analyzing, I notice that the average sequencing depth on Y chromosome vary a lot more than X chromosome, and the depth is mostly less than the average depth on autosomes (see the figure for an example; from Turro et al. 2020). Most individuals that got sampled do not have diseases involving large deletion of Y chromosome.

What causes the variability in the sequencing depth on Y chromosome? Why is the average depth on Y almost always less than that on autosomes?

[German.M.Demidov]

Hi, there is a pseudoautosomal region on chrX/chrY which is normally excluded from calling and it affects a large part of Y and not so large part of X. Also "an average coverage" is more variable for smaller chromosomes.

Also XYY caryotype is not that rare.

https://en.wikipedia.org/wiki/XYY_syndrome

Also it depends on reads with which minimum mappability you count

Inconsistent Coverage on X chromosome between genders

[samuelandjw]

We usually exclude reads mapped to PAR before calculating average depths on X or Y chromosomes, so I think PAR reads will not affect this much. XYY karyotype will increase the depth on Y chromosome significantly, but as I mention, most of males with no abnormalities in sex chromosome copy number will still have highly variable Y sequencing depth.

[German.M.Demidov]

tbh I do work with exomes/genomes a lot and I wrote a CNV caller which infers sex based on X/Y ratio ( https://github.com/imgag/ClinCNV ) - but Y coverage is not noisy there at all

So maybe the data that you showed were under-corrected

[crisime]

Hi German,

Did your test-cases include older men? (See my answer below) We use your tool and ngs-bits and get quite a lot of men 60+ in the unknown area when checking Y/X-ratio.

Thanks for your great work!

Chris

[German.M.Demidov]

Hi Chris,

absolutely, I would say >30K people were clinically tested with ClinCNV and there were older men. There were many problems with the karyotype, but I can't say that any of them were much prevalent. Old women rarely loose (in mosaic) X chromosome, but males normally don't loose Y.. This is very interesting. I assume maybe these are cancer patients and you sequence blood after chemo?

[crisime]

I never checked for cancer in our patients documented history because the high age explained it in accordance with the literature.

For example, using a threshold of ≥10% of affected
cells in three independent cohorts studied by SNP arrays,
a recent study demonstrated that about 7, 14, 18, and 20% of men were affected in the ages of ≤65, 66–75, 76–85,
and ≥86, respectively (Dumanski et al. 2016). These data
show that LOY accumulates with age in the studied cohorts and that LOY is the most common acquired human muta- tion during life in normal peripheral blood of men, and it
is affecting ~1.6% of the human genome (Forsberg et al.
2017).

Loss of chromosome Y (LOY) in blood cells is associated with increased risk for disease and mortality in aging men

LOY is seen as risk factor for cancer, but not the other way around.

Mosaic loss of chromosome Y in peripheral blood is associated with shorter survival and higher risk of cancer

It is strange that you don't see any of this in such a large cohort.

[German.M.Demidov]

It also depends which degree of mosaicism we'd check - I normally check anything happened in >20% of cells. Otherwise it becomes much more tricky to detect.

20% is a bit too much I'd say in >86yo. It is certainly a part of clonal hematopoiesis and the frequency of all CHIP events is around 20%. Tough to believe that all males with CHIP have Y loss.

https://www.nejm.org/doi/full/10.1056/NEJMoa1408617

[crisime]

The paper you shared is detecting CHIP events (SNP/indel) in 18,4% of >90 year old people by looking at just 160 WES genes. So it does not make any estimation for the frequency of all somatic mutation events or anything at the level of whole chromosomes. Or did I miss anything?

[German.M.Demidov]

Correct - could be more frequent for aneuploidies

https://www.nature.com/articles/s41586-020-2426-2

[samuelandjw]

I added a figure for Y/Auto depth as a function of age of one of our cohorts. Is the trend similar to what you observed in your data?

[crisime]

I never made a figure. We just check the gonosomes of our patients for the ratio of reads on Y/X, to detect special cases of gonosome distribution. The only cases triggering our threshhold for lower than normal Y-fraction apart from Klinefelter-Syndrom (XXY), have been 60+, most of them even 80+. So I never saw the whole picture of my data, but I can confirm the lower right part of your figure ;)

Answer 1

[crisime]

Hi Samuel,

I would guess it is the age related "Loss of the Y chromosome" you see in the data, which is a known effect. See: Loss of Chromosome Y and Its Potential Applications as Biomarker in Health and Forensic Sciences for example.

"Loss of chromosome Y (LOY) is a mosaic aneuploidy that can be detected mainly in blood samples of male individuals. Usually, LOY occurrence increases with chronological age in healthy men."

"The mechanism by which LOY occurs is yet to be elucidated. It is not known whether the Y chromosome is lost as a whole or if it is a gradual deletion process during the lifespan."

We see that quite a lot in our WES data of older men.

[samuelandjw]

Interesting. Today I analyzed the WGS sequencing depth on Y chromosome in a cohort with male samples of 60+ age and the depths are systematically lower than what I have seen before in other cohorts with younger male samples.