There's also experimentation to generate hypotheses. I.e., let's sequence the genomes of healthy people and the genomes of people with some disease and see if t

There's also experimentation to generate hypotheses. I.e., let's sequence the genomes of healthy people and the genomes of people with some disease and see if there is a difference. It's possible to argue that the implicit hypothesis there is that there is some sort of difference, but that is a particularly weak hypothesis compared to the specificity of hypotheses that can be formed once we have this pile of data -- i.e maybe X, Y, Z differences are important in this disease.