I think the initial tests would be computer models. Take models of how the earth's core and mantle work (or the Oort cloud) and intersect that with models of how dark matter is spread through the galaxy. Are there inputs where the models behave as predicted, and are the inputs in any way realistic?