I think people trying to do these analyses are fooling themselves. The data we have and understanding for the vast majority of variants and combinations of variants is insufficient to produce accurate explanations of our current health or make predictions for the future.

I started working in this field many decades ago and it's taken me quite some time to see just how dysfunctional the literature has become in terms of describing things that are useful to individuals. Our understanding of genomes, genotypes, phenotypes, and their relationships, is incredibly limited, but you wouldn't know it when reading a bioinformatics paper.