It's not particularly difficult to determine the protein that a bit of DNA codes for. It's more difficult to figure out which of those are going to be reasonable antigens to target, but you don't really have to. Cancer cells aren't unknown pathogens, they're regular old human cells with mutations.
You don't need to do that either though. Cancer mutations aren't infinitely diverse. "Personalized medicine" sounds like a treatment just for you and you alone, and maybe in a Star Trek future it will be, but in the meantime it means a targeted treatment. You'd identify something that occurs in 10% or 1% or 0.01% of a particular type of cancers, make a treatment, and sell that along with a test for that mutation. We've already got several of those based on more traditional immunotherapy. RNA vaccines just make it a lot easier so we'll have lot more options, including ones that target the 1% and 0.01% instead of just the 10%.