I worked in a wetlab or for a bioinformatics company writing software for 4 years. One of the things that completely shocked me when I started working there, and I think most people don't appreciate, is that taking a liquid and figuring out what proteins are in there is a very hard thing to do. Then the next step -- how much does my sample have of the things I've identified -- is also hard to do. Doing either of them inexpensively is harder still. People still work hard on those two questions today, and there is still room for significant innovation around those two questions.

Also, CSI style science, where you drop a sample into the magic machine and an answer pops out is... nonsense. These are some of the most finicky machines you've ever encountered -- you always end up doing baselining / calibration runs, etc. Often they involve skilled lab work to set up, calibrate, run, and determine whether answers are significant.