You can't do intelligent drug design if you don't know what the target protein looks like. We've gotten great at solving protein structures with things like crystallography and cryo-EM microscopy. Unfortunately, many interesting drug targets reside in the membrane of a cell, which means you can't easily work with them in a lab because they aren't soluble in anything but a plasma membrane. For instance, this is an issue with the 5HT2A protein, a g coupled protein receptor that is implicated in many serotonin related pathways.

Being able to predict what it would look like would be a huge deal because then you can go about intelligently designing drugs for it.