Hi Wei,
In our experience this can be caused by having an incorrect cif file for the ligand, especially if you see it fits well in Coot but looks wrong after refinement. It's a pain to go through the cif file by hand, but you can read through it and see if all the restraints are correct, or take a close look at the geometry of the refined ligand. You might be to try generating the cif file with ProDRG and the PHENIX tool, and compare refinement done with each. Of course, the easiest solution is to take up Pavel on his offer to take a look at it! He and others on the PHENIX team have helped us quickly resolve similar problems with the ligands a number of times, and we ended up on a paper together on one particularly nasty case that he spent more time on.
all the best,
Kendall
On Oct 24, 2013, at 10:19 PM, Wei Shi
Hi all, I am working on a structure of protein-ligand complex. Four ligands are placed for the dimer protein and the density for the two ligands of the first monomer is better than the density for the other two ligands of the second monomer. Ligand is moved to fit density better in Coot (and for two ligands of the first monomer, they fits the density almost perfectly), but after a refinement in phenix (default settings + NCS restraints + Secondary structural restraints + Optimize X-ray/stereochemistry weight +Optimize X-ray/ADP weight), some part of the ligand which fits density good before moved out of the density again...Besides, it always shows green density in Coot for the ligand region even in places where the ligand is in density... Any suggestions or ideas about how to fit the ligand better and why the density for ligands of the second monomer is worse than that for the first monomer and why the ligands would move out of density after refinement? Is it because the protein model is not good enough to get the ligand density good? There is a conformational change upon ligand binding, and I rebuilt some part of the protein manually, and the density for a region of about 30 residues is not very good, and I tried to mutate those to alanies and refine, but it didn't help me see the density better... Any suggestions or ideas on how to improve this protein-complex structural model? Thank you so much! The statistics for the current best model is as follows, and the resolution of the dataset is 2.8Å.
start final ------------------------------ ---------------- R-work: 0.3359 0.2993 R-free: 0.3619 0.3558 RMS(angles): 1.03 1.55 RMS(bonds): 0.006 0.007
MolProbity validation Ramachandran outliers: 4.7% (Goal: < 0.2%) Ramachandran favored: 85.3% (Goal: > 98%) Rotamer outliers: 4.5% (Goal: 1%) C-beta outliers: 0 (Goal: 0) Clashscore: 7.43 Overall score: 2.56
Thank you so much!
Best, Wei _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb