Got a full volume in Local refinement

Hi all,
I’m tried to refine a mobile domain of my protein by 2 ways.
Way1, by directly running local refinement with a mask of this mobile domain. However, the result was still my full protein volume. Why does this happen?
Way 2, I did local refinement with particle subtraction first, and I got a this domain’s local volume, but result seems weird(see the below picture). By the way, this domain is more than 100KDa.

What should I do to improve this domain’s resolution?


I think strange result… you have focused the alignment process on that domain, but full particle is still there in the first case, but the mask should remove outside density. double check that subvolume mask is what you put in static mask parameter. after subtraction, the info outside the mask is removed so should be mostly gone after the refinement. Now what IS expected is focused 3D classification should have motion within mask separated to classes but region outside of mask is same for all.

your results don’t look good, but also not atypical. this job type is hard to get success. try dynamic vs static mask. first make sure the parts you are trying to align are as close as possible. try refinement with same mask prior to local refinement with mask, if domain is big enough (100kda is on the lower limit, prob wont work, maybe the signal subtracted particles will have better success - in part this depends on featurefulness). increase or decrease the shifts and rotations sampling frequency and range. recentering per iteration or not. generate several masks and try try try. a good mask has extended edge, and is soft, but still contains only the part which is mobile together and NONE of the outside part that is rigid. Use ctrl+shift selection in Chimera to select a domain boundary instead of using sequence to select.

1 Like

Thanks a lot first!
In the first case(local refinement with a mask), I double checked my input parameter, my mask was my interested domain and the mask parameter was “static”. I have also done a non-uniform refinement with this same mask, I got a correct sub-volume of my mask, but the resolution was not improved.
For the local refinement with particle subtraction, I will try your advices. One more thing to confirm that the mask should only contain the mobile part ? I usually used a mask contain a small part of my rigid domain and all of the mobile domain to make sure my mobile part in correct location in model building. If the map of local refinement is all mobile , How can I know my interaction interface was right in model building?


There is no perfect place to align it, since it is mobile. But it is sufficient to build it independently and then align it to the consensus map (whole particle reference). It’s a bit imperfect to suggest any local refinement can be connected to the full map. Just have to be clear in methods and put local refinement map online.

First problem first though, get a good structure for region or interest, and my best success is excluding the rigid part

Got it. Thanks!I will start from different masks.