Nice work, Adam!!! Great to see that what we observed on a public mouse Dmnt1 KO dataset replicates on human cell lines as well!!! 🙌🙌

Overall, we see that, in the UK Biobank, individuals with higher rDNA CN tend to have blood cell composition akin to systemic inflammation states and worse kidney function. So, to answer our original question, yes! It does matter how many copies one has! (9/9)

When the second release of sequencing data became available, with ~200,000 new white British individuals, we repeated our analyses and we were delighted to see all key results replicate! (8/9)

And blood composition changes are not the only consequence of high rDNA CN we observe. Biomarkers related with kidney function (such as the eGFR ‑ estimated Glomerular Filtration Rate) worsen as rDNA CN increases, leading to a higher risk of Kidney Failure (7/9)

Confounding also doesn't seem to be the culprit. When using stringent subpopulation filters and/or controlling for potentially-confounding variables, associations remain! We also see similar results in non-white British individuals (6/9)

But could differences in blood composition be driving the differences in observed CN, and not the other way around? In situations we know NLR changes, we see no change in rDNA CN — such across sexes, with ageing, in diseases, or throughout the year or time of the day (5/9)

Does this variation matter? Associating rDNA CN with thousands of phenotypes revealed blood composition measurements often significantly change with increased CN, particularly in markers of systemic inflammation (such as NLR — the neutrophil-to-lymphocyte ratio) (4/9)

From the first data release for >150,000 white British participants, comparing rDNA CN of relatives resembled what we would expect from anywhere else in the genome. What we didn't see is any evidence of variation elsewhere in the genome driving rDNA CN (3/9)

The UK Biobank offers sequencing and phenotypic data for 100s of 1000s of people. So we were thrilled to realise we could derive a proxy for rDNA Copy Number (CN) from the available alignments. But a warning: where the data was generated must be taken into account! (2/9)