I think my protein tetramer is pretty big, but why can’t I get high resolution in 2D classification when picking particles? I ran 40 iterations and classified into 100 classes.
you definitely have compositionally heterogeneous data - there are big, small, and smaller particles. try 200 classes, and increase batchsize to 1000 (this will make the job run pretty slow) for 20 iterations. from the first 2D, use large classes to repick particles. and use small stuff to repick particles. see what comes from each. when running “inspect picks”, does it seem like particles are well-picked? your contrast and defocus look great, but ice quality can always be an issue. my guess is 1000 movies is not nearly enough data to make this negative conclusion.