Pro:
* Learning about genomics, immunology & cancer
* Working on clinical trials (i.e. might help a real person)
* Seeing how science works as an institution (interesting peek inside the sausage factory)
Con:
* Doctors (& some biologists) tend to be more hierarchical than I'm used to from CS. You sometimes need to flash credentials to be treated respectfully (I have a PhD in CS, so that helps).
* Hospitals and their bureaucracies are insane. So, if you want to actually have patient impact you'll have to deal with some really byzantine processes (and the madmen who create them).
* Bioinformatics tools are crappy and their authors have few incentives to learn software carpentry. Related: file formats are designed to solve a specific problem and then get sloppily extended by everyone else.
* Biology accumulates a lot of historical baggage in its naming conventions (i.e. weird names are often explained by "SoAndSo thought it interacted with SomeOtherProtein in 1953"). This makes learning a lot harder than it could be.
The reason for moving away from software was that it's vanishingly rare to do important work without expertise away from your laptop. As much as startups like to say they're "changing the world", I felt that most meaningful changes are simply leagues beyond the kind of complexity you can solve with coding.