Using the extracted particles, we run 2D classification jobs in cryoSparc, using 40 iterations to clean our particle dataset. Selected particles then were use to train a Topaz model, and model was used to pick particles, followed by rounds of 2D classification. Selected particles can then be re-extracted to original pixel size (in this case 0.935Å, box size of 320pixel for monomers and 400pixel for dimers)