Tips to refine flexible dimer domains identified by 3DVA


I have a very heterogeneous sample with monomer and dimer particles. One domain of each monomer interacts to form a 4-domain dimer.

I have performed 3DVA on the 123k dimer particles (filter of 10 with 5 modes seems optimal) and it shows rather large swinging motions of the two free dimer domains (poorly resolved after NU refine) while the domains involved in dimerization are very stable and only show slight breathing.

By performing signal subtraction and local refinement I have obtained a good reconstruction of the interacting domains but can’t figure out how to separate particles for the different states of the full dimer. Obtaining the full dimer would be very informative w.r.t. the mechanism of physiological processes so I would appreciate any suggestions as to how I could proceed.

What I have tried so far was 1) 3DVA display clustering with 4 clusters, 2) heterogeneous refinement with 3 classes, 3) 3D classification w/o alignment in Relion (using T=40, K=8 and either a wide 4-domain mask around the full dimer or a 3-domain mask excluding the more flexible left-hand domain). None of these approaches gave good separation into different 4-domain states. Can multi-body refinement be applied in a case like this? The monomer is only 140kDa and I think the individual floppy domains would be too small to define as separate bodies.

You could try using the volumes from 3DVA that interest you, and use them as input of heterogeneous refinement.

Thanks @mmiotto
I will try that :slightly_smiling_face: