Trouble fitting a complex into <4 Å map - severe clashes at interface despite tilted dataset

Cryo-EM Data Processing
1

Hi everyone,

I’m working on a Cryo-EM reconstruction of a T-cell receptor bound to a peptide-MHC complex. The final map is at ~3.7 Å (no imposed symmetry), reconstructed in CryoSPARC from a dataset that includes both untilted and 30° tilted micrographs.

Despite the improved orientation distribution, the fitted atomic model shows major clashes at the TCR-peptide-MHC interface (compressed along the y-axis) - particularly between the CDR loops and the α-helices of the MHC. The overall fit to density looks reasonable, but the interface region is very strained even after real-space refinement in Phenix. The pixel size is not the issue. I have also tried global and local CTF refinements. This issue appears for a few but not all TCR-MHCs we study (no clue what the relationship is). Any advice on how to diagnose and correct this would be much appreciated!

Best,

Marta

2

Hi @mtborowska

Did you try sharpening tools like deepEmhancer or EMready before you fit your model into the map?

3

deepEMhancer yes. I should also clarify that adding tilt data improves the complex interface overall and reduces the clashes, but it does not resolve the issue completely.

4

@mtborowska try phenix reduce and hydrogenate your model and then perform realspace refine. After the clashes comes down, you can remove hydrogens.

hope that helps.