Streaks and artifacts in 3D classification but great 2Ds


I have been having an issue generating a 3D reconstruction of a >100kDa membrane protein. The 2D averages look great with clear secondary strucure in both top and side views, but the 3D volume shows streaks in the top view and from the side view appears to show a horizontal planes where the density is weak. Any ideas about what might be going on? I imposed C4 symmetry here but see a similar effect even with C1 symmetry.

Nice 2Ds! What parameters did you use for ab initio and how does the resulting map look? For something like this (aquaporin?) with not much outside the membrane, I would recommend starting at relatively high resolution - e.g. an initial res of 8 and final res of 6, something like that.

Yeah, it’s an aquaporin. I tried a couple of different initial models - one generated from this dataset is shown below and one form another of the same protein that looked fine in 3D classification. I used default settings other than C4 symmetry (initial 35A, max 12). I will try generating an initial model with a higher resolution, though when I used a map from another dataset of the same protein that didn’t have this issue but even with that initial model I saw the same behavior.

I would also try ab initio in C1 - you shouldn’t need to enforce symmetry in ab initio (occasionally necessary, but I wouldn’t do so unless you find that C1 doesn’t work, and enforcing C4 in ab initio can give artifactual densities which could propagate to downstream refinement, even if symmetry isn’t enforced in the subsequent refinement)

Thanks for the advice! I will give it a shot.

Out of curiosity, is this EMPIAR-11592? I experienced the same issue reprocessing the dataset–back in a different life–but didn’t try too hard to figure it out. I don’t get to see cases of symmetry in my day job. Please do share updates when you get further.