But don't forget that these methods use class-averaging and each class represents members of a frozen structure. One of the beauties of cryoEM is that if your s

But don't forget that these methods use class-averaging and each class represents members of a frozen structure. One of the beauties of cryoEM is that if your structure is dynamic, cryoEM will actually capture the ensemble in each class, which if you have enough images of each class you can solve each states' structure. One caveat is that if there are no local minima, you may have a huge number of states which means you won't get high resolution structures of each, but typically you'll have a few conformations that are reasonably stable. The other cool thing is you can throw in literal cell lysate and let the computers "purify" the sample for you. This allows the native environment to participate in the thermodynamic landscape of dynamic proteins, possibly showing more meaningful dynamic states.