I have a data set with overall high resolution but struggle with resolving a small subunit that seems to be flexible and thus not present in the consensus map.
Specifically we are talking about one of the lobes of one calmodulin, meaning approx 8-9kDa, which is not present in the map.
However, I did 3D classification with focus mask on the area where I assume the lobe to be. This resulted in a class that actually shows some mass in that region. 2D classification also shows a small amount of images with mass in that area.
Local refinement wasn’t too successful yet, I’m afraid it’s just too small? I manually set the fulcrum and tried to restrict the angular and translation search as well as used a mask with soft edges.
What are your suggestions? Could 3D Flex be an option if I carefully pick only the class that contains mass in that area?
CaM and alike proteins are intrinsically very flexible. It is not unusual to miss an entire lobe, even in crystal structures. And even when you see the entire CaM, map quality is usually worse than for the partner protein. Now in SPA what has worked best for me was 3DFlex with a custom mesh. Don’t make too many subdomains, just focus on the parts that move away from each other. And this you’ll know by running a preliminary 3DFlex job with the standard (single) mesh. And don’t expect miracles either. Be happy if you can place a few helices.
I also believe local refinement can not deal with such small bits - at least I’ve never any good results from it in cases like this.
Thanks, 3D Flex is what I’m working on at the moment. Often the 1st CaM is a bit more defined, as it interacts with the head. One of the lobes is actually quite well resolved in our structure, the second one however almost completely missing due to flexibility