With enough data it's possible to disambiguate between genetically linked behaviors and learned behaviors because of the different diffuse patterns through disparate populations. The summary is that behavior and intelligence is absolutely dominated by genes, there is really only a 13% residual - so on a nature vs nurture spectrum the nurture is less than 13%.

The stronger the selection criteria bias on intelligence the more likely a RCCX gene variant (SNPs) will be selected, specifically TNXB (burnout / ME/CFS), C4 (schizophrenia), and CYP21A2 (generalized anxiety disorder).

You may wonder why this doesn't come up in GWAS studies - usually there is an assumption of independence of SNPs due to the idea that most SNPs are very old and genetic drift has been sufficient to provide this independence. Since these individual SNPs are so behaviorally impactful there a strong sexual selection bias that takes place so those assumptions don't hold and thus the math is bunk. I could spend forever talking about the incompetence of medical research.

Dr Jessica Eccles is probably doing the best research where she talks about the shocking 'overrepresentation' of Generalized Hypermobility Disorders (GHD) in the Long Covid populations. In my view the only difference between GHD and hEDS is the number and severity of TNXB SNPs, so while 2% of the population are impacted most of those are only mildly impacted. The co-occurrence of multiple TNXB SNPs is much higher random chance - again from the aforementioned sexual selection biases. In general researchers into such conditions assume TNXB SNPs are far too common to cause such a rare disease, the thing is that it's not a rare disease - just rarely diagnosed and surfaces in a variety of complex ways that look completely different to the untrained eye.

You can get a Deep Whole Genome Sequence for ~ $280 from dantelabs and then you'll know for sure.