Thanks so much Pavel ! I already went through placing the crystal structures manually in the density map and building the entire model. However, I didn't use the reference model restrains during the refinement and that explains why every time I refine, the geometry goes bad. I'll do this as you suggested. Best, Ashu On Tue, Nov 6, 2018 at 3:47 AM Pavel Afonine wrote:

Hi Ashu,

you can use phenix.dock_in_map to dock known pieces (from crystal structures) into cryo-EM map.

Then you can use phenix.real_space_refine to refine placed model into cryo-EM map. It is important that you use reference model restraints with high-res X-ray model as a reference.

What to do with side chains is a good question. I doubt you can see any side chains in 7A resolution map. Reference X-ray model is perhaps your best source of knowledge about the side chains. Though keeping them in the model in the absence of support from the data seems questionable. I doubt there is a clear cut consensus/answer to this.

You can vary the strength of reference model restraints do vary the dose of bias from the known (reference) model.

Pavel

...

Dear All,

I am refining a ~7A single particle cryo-em structure of a protein. The protein has two domains and high-resolution crystal structures of individual domains are already known. I could fit these two crystal structures in the low resolution cryoem map of the full-length protein. I have two questions:

1) Is it acceptable to refine with side chains in this case? In most of the cases, with the guidane from crystal structures, side chains could be placed with reasonable confidence, particularly smaller ones. Should I restrict to poly-ala model or take the advantage of crystal structures and include side chains also?

2) When I tried to refine with side chains, longer side chains like K, R, Q, E which have weaker density tend to bend towards the backbone density. Is there a way to prevent this happening?

Suggestions/comments from the community would be highly appreciated.

Thanks !

Ashu