What can I do if the density map of the regions in motion from the 3DVA cluster mode is too bad to be properly builded?

Hi, all

I used the cluster mode in 3DVA to separate different conformations of my protein.
But the density of the motion/flexible region is still badly resolved and the model cannot be properly builded.
I’ve tried low-pass filter between 5 to 15 A when running 3DVA.
However, that region still cannot be resolved clearly enough to do the model building.
Also the local refinement didn’t work.
I used the same number of low-pass filter for 3DVA and 3DVA display.
Will it help, for example, low-pass filter at 10 A for 3DVA and display without low-pass filter ?
Generally, should the low-pass filter be the same for 3DVA and 3DVA display?

Because the region accounts for nearly half of my protein, I cannot just ignore that region during model building.
Is there any suggestion?

All the best,

Hi Jessica,

What is your workflow prior to 3D-VA? Are you doing local refinement of the mobile region, then 3D-VA, or are doing 3D-VA starting from the global consensus refinement? What resolution is the preceding refinement?


Hi Oli,

I did 4 rounds 2D classification and a heterogeneous refinement to remove junk particles first, and then I did a global Non-uniform refinement before 3DVA. The resolution of NU is 2.59 A.
I did the local refinement after 3DVA display using the cluster mode.

The resolution of other region is pretty good, but only the flexible part cannot be well resolved.
Should I do the local refinement before 3DVA?

Hi Jessica,

Yes, I would definitely try doing local refinement first, then 3D-VA.

If your global refinement is 2.6, then I would also try using higher resolution for 3D-VA - even if it is very flexible, sometimes a filter res close to the global res will in our experience give more stable results.

Good luck!


Hi, Oli
Thanks for your reply, I will try to do a local refinement before 3DVA and use a low-pass filter close to the global refinement.

I have a question about local refinement in cryoSPARC.
I thought the local refinement job (new) should generate a sub-volume of the region covered by a mask, but it generated a full volume of my protein. Is there something wrong? Is it already combined and can be directly used? or should I still combined it using vop maximum in Chimera?


The mask used for refinement will limit alignment to this region, but the output volumes are not masked. So you expect to see stronger density in the region within the mask (because that is the region you are aligning on), and weaker density outside. Is that what you see?

So the alignment is already done in the local refinement job, I can directly use the output volume for next job. Do I understand correctly.

Hi, Oli

The input for 3DVA is only particles and a mask.
I don’t know how can a local refinement before 3DVA be helpful.
These two jobs seem to be unable to be connected.

Also, I compared the volumes from local refinement and NU refinement, I think volume from local refinement should be combined with the NU refinement, or the resolution outside the mask become worse.

Hi Jessica,

Use the particle set output by local refinement for 3D-VA. As they are better aligned (less variability) in the masked region, this can sometimes give better results in 3D-VA.

This is expected. If you are aligning on a mobile region, improvements in that region will come at the expense of degrading density outside the mask. Creating a composite map is one way to approach this.

Got it.
Thank you, Oli.

Is there any way in cryoSPARC can make a composite map?
Or I just use Chimera to make it and input a 3D volume into cryoSPARC?

Hi Jessica,

Making a composite map is only for display, really, and for convenience when fitting/interpreting an atomic model. That all happens outside of cryoSPARC, after you have finished processing each individual region.


I see.

Thanks! :smiley: