Symmetry

[mcianfrocco]

Hi @zhonge -

How do you recommend we prepare data from symmetric particles? Should we run symmetric refinement (e.g. D2) or an asymmetric refinement (C1 symmetry) as the consensus refinement? And, if we are doing a C1 refinement, should we 'symmetry expand' the data so that all asymmetric units are oriented in the same manner?

Thanks!
Mike

[zhonge]

Thanks for the question! I would recommend a C1 refinement and to start the refinement from an asymmetric initial model. The goal is to get the true pose for each particle, and I’ve found that, in practice, using an asymmetric initial model better ensures that each particle is aligned to the correct symmetry copy. Otherwise any asymmetric motions/heterogeneity would be divided between the symmetry copies, and have to be modeled separately in the latent space.

I would recommend against symmetry expansion. If I understand correctly, symmetry expansion locally aligns N copies of each particle to the asymmetric unit, which can give a better static structure for the asymmetric unit, with the remainder of the complex masked out. It could make sense do in some cases [1], but given that there are many types of possible "pseudo"-symmetries, I think it would be easiest to just treat the complex as C1 as a first pass.

[1] If the heterogeneity is expected to be purely within the asymmetric unit and independent of which symmetry copy it is present in, symmetry expansion + signal subtraction could make sense, especially if the extra particles are needed to get a higher quality map.

[zhonge]

As a follow up, it may be important to change any dynamic masking parameters so that the mask used in the C1 refinement includes the asymmetric/heterogeneous regions. Otherwise, the effect will be to only align on the symmetric region, leading to inaccuracies in pose estimation.