My thought process here is that perhaps these trans-pQTL hotspots are reflecting some biological factor that is systematically affecting the binding chemistry across a large fraction of binding sites?

Quite possibly power - but then some of the trans-pQTL hotspots have really small p-values across many proteins (e.g. the APOE/NLRP12 locus on chromosme 19) in Sun et al 2018, which is only 2x the sample size here.

I'm quite struck by Figure 1 - there don't appear to be strong trans-pQTL hotspots like we typically see with affinity-based technologies. This makes me wonder if previous trans-pQTL hotspots reflect systematic epitope effects - is this something you've explored at all?

prcomp() uses BLAS, which depending on the BLAS library and how R was compiled against it when R was installed will default to using all available cores.

Try adding the following code before your call to prcomp():

library(RhpcBLASctl)
blas_set_num_threads(1)
omp_set_num_threads(1)

Reminds me of this classic tweet:

GPT has consumed too much assassins creed lore: assassinscreed.fandom.com/wiki/Genetic...

Now's a good time for a reminder about the internet wayback machine: web.archive.org

With accompanying editorial piece summarising and contextualising these findings written by myself: www.nature.com/articles/s44...