I am attempting to perform a 3D reconstruction of a low pH-treated icosahedral virus, but I’m struggling to extract sufficient information from my current results. I have approximately 37,500 particles, which yield a few good 2D classes upon classification. However, I consistently fail to obtain a suitable initial model for reconstruction. This issue persists regardless of the particle group selected or attempts to enforce icosahedral (I) symmetry during ab initio reconstruction. Increasing 2D class sizes to 100 and adjusting O-EM iterations to 40-80 have not yielded a good class for an initial model. What alternative approaches can I try?
What settings have you used for ab initio? I find for high symmetry, turning off “Centre structures in real space” is critical. Leave at C1 imposed. Then if you get something which looks “OK-ish” in one or two directions, use Volume Alignment Tools to align to I1. Then do a homogeneous refinement with the VATs-aligned initial model (turn off symmetry alignment in the refinement job!) and symmetry imposed.
That initial model looks like it might be a reasonable starting point, though. Try aligning to symmetry with VATs and a homogeneous refinement, see what comes out the other side…
1 Like