Combine information of homodimers with monomers


is there a way to combine the whole information of picked homodimers with picked monomers?

My example with a not perfect dataset:
I have a 170 kDa protein and picking mostly monomers, 90K particles in the end delivered 3.7 Å structure.
A parallel picking of homodimers delivered 4.3 Å with 43K particles. In C2 symmetry it goes down to 3.8 Å. Dimer picks have more other views.

I combined both particles classes and masked one monomer out: 133K particles delivered 3.4 Å. Masking worked not perfect, still some density for the other unit. Played around with the masking settings.

So, is there a way (e.g. with symmetry expansion), do use the information of both units of this homodimer and combine it with the monomers?

Would double theoretically the information of the dimer to 86K monomer particles and in total 176K particles. Did not found a way. Thought could be maybe possible to deliver some vectors/distances about the orientation of the units, extract the position of the units new and recenter only to one part. Of course will not be perfect because of the close density of the other unit but this effect do I have in the moment with the masking also.

Thanks a lot for your input.

I only missing maybe 0.2 Å for the ligand binding question I have.


Hi Moritz

Yes - symmetry expansion should do the trick here.

Just to clarify, do you only seek to perform a masked refinement of both protomers of the homodimer, or you want to do that and also combine with the monomer class?

In the former case it is straightforward - refine in C2 (assuming your homodimer is C2 symmetric), perform symmetry expansion (C2), then run local refinement with a mask around one protomer.

If you then want to combine with the monomer, it might be a bit trickier but possible - you can re-extract after symmetry expansion, centered on one protomer (use volume alignment tools for this), and perhaps perform signal subtraction to reduce the contribution of the other protomer.


1 Like

Thank you Oli,

this sound good.

Yes, the second one is the goal.
I am a bit lost with the “centered on one protomer (use volume alignment tools for this)”.
Do I have to set “3D Coordinates of new center (voxels)” and do this two times for each protomer?
Input would be the volume of a homodimer and mask a monomer? … I will try to find some help in the Guide…

You only need to do it once - symmetry expansion helps you there.

Symmetry expand, local refinement (with mask around one protomer), then volume alignment tools (on the coords of the center of the protomer mask, which you caj get using measure center in Chimera).

You can then re-extract the “subparticles” (which will now be centered on each protomer) using the output of volume alignment tools.

Then you can combine that with the monomer data (assuming the same box size etc) and perform a consensus refinement. If in the dimer the density of the other protomer causes problems in the consensus, you may want to subtract it prior to the consensus refinement. Does that make sense? Sorry it is a little bit convoluted!

Thanks Oli,

I am trying to fight my way through it.

  1. local refinement (with mask around one protomer): Input: Dimer particles after Sym Expand in C2, Monomer volume, Monomer Mask → does not deliver well, but not the final goal

  2. then volume alignment tools: no symmetry alignment, Input: Dimer volume, Monomer mask, Dimer particles after Sym Expand in C2, Chimera center of monomer mask
    → extraction from micrographs is running, it looks like there is a shift between the two extracted particles

still playing around.