If you make a connection between genes and disease, it is often a complex/weak link. Most of the on/off genes have already been discovered, by now we have gotten down to the more complex ones based on interactions, etc.

Having worked with genetic data a lot I know that the reason for the wealth of European genetic data therefore lies in the limited generalizability based on more diverse datasets.

If you try to generalize from a worldwide genetic dataset, you will probably fail. Your inferences from this diverse dataset to individual genomes will be abysmal. Predictions are already difficult to make from a country-specific dataset to an inhabitant of such country.

I don't know why people always want to see race everywhere. For scientific reasons, it has been way more convenient to sequence more people of European origin. In my opinion, that has nothing to do with racism. It is a matter of what data has better availability and is more valuable right now (not even in terms of application - simply in terms of being able to detect complex underlying patterns). Now that the cost of sequencing is getting lower and lower and availability increases everywhere, it is great to see more diverse genomes sequenced. This will help us understand more of the genome - still, this is no magic bullet by a long shot. The problems stated above get bigger with dissimilarity. Many discoveries made in African genomes can probably not be generalized unto the remaining world.

Current methods are just not sufficient to dig deep into the genome and its mechanics. At this point, we would probably need to sequence the whole of earths population a few times to be able to make consistently accurate and significant predictions across many complex traits and diseases.