Hey @hansenbry,
Due to the ambiguities in helical symmetry that are inherent in the way projection images are formed, it can happen that the asymmetric refinement doesn’t converge to a solution for the symmetry parameters, and thus the correct solution doesn’t show up as the global optima in the table printed out (or even that the wrong solution is converged to). Unfortunately, this is why asymmetric refinements and ab-initio reconstructions with helical symmetry are not guaranteed to lock on to the right symmetry, and it can depend on the dataset and initialization. We still chose to illustrate a workflow in the tutorial where asymmetric refinements/ab-initio reconstructions can work, because for many datasets this is a viable workflow (especially filaments with large asymmetric units, like EMPIAR-10495 and EMPIAR-10213). There’s a bit more of a discussion on this phenomenon in the Helical ambiguities post, and on Prof. Egelman’s paper on the subject.
In your case, it looks like the 15th row is fairly close to the true solution, with a rise of 5.051 Å and twist of -103.7º. You could try doing a second refinement with these symmetry parameters, which are a lot closer to the true optimum.
15 | 1.753e+01 | 3.471e+00 | 5.051e+00 | -1.037e+02 | 2.771e+03
Best,
Michael