Genetics and Epigenetics research at scale on humans. Just getting the database together in a machine readable form would be a huge accomplishment
There is a huge (missing) sample problem in genetics. This creates problems in epigenetic research, even with gene variations where there are accurate sample sizes, because it becomes extremely difficult to figure out if it is the gene that it is the trigger or an epigentic factor turning on a gene, and if so, what sort of epigenetic factors are we looking at in the first place. Turning off genes through behavior changes is mostly off the plate without getting accurate data first.
Part of the reason: It is still expensive to decode every genome in a human (around of $250k per person right now) That cost will drop radically in the next 10 years because it is only 1-5 parts that really need to drop in cost to make it around $250/person. Then matching that data to health and other related records en mass is huge- especially because most basic health records in EHR are geared towards billing procedures and short term case discussion, not tracking people, families, and communities.
Furthermore, it is a true ethics issue. To donate part of your full genome to science means parts of your genome will be bought and sold for research purposes. The going rate for simple stuff is about 1k/genome, with more money to the market with more interesting medical histories and family data attached. Creating a really big database that is open means this data stops being in a market and could push for innovation in the field.
I'd like to remind you that we have no idea why the county that Ycombinator sits in is among the top counties in the country for Breast Cancer rates. Just getting an accurate sample size of Marin, San Mateo, and San Fransicio counties's genetics and true medical history could make a huge impact on why people get cancer period, because we have no idea why there is a cluster there.