I frequently utilize Phenix.phaser and Phenix.refine, along with CCP4i, to address these issues. Despite numerous attempts through trial and error, I have not yet formulated a definitive protocol or conclusion to determine the optimal refinement strategies for resolving the problems in both initial and subsequent refinement stages.
Should I tailor the refinement strategies based on the resolution of my protein crystal's diffraction pattern? Alternatively, is the most effective approach to focus on obtaining high-quality crystals and improving the diffraction data to achieve a resolution better than 2.5 Å.
I appreciate everyone's responses in advance. If any of you could provide me with some keywords and reference materials for me to explore on my own, that would be greatly appreciated. I apologize for my limited knowledge and English proficiency, which make it challenging for me to comprehend and learn about these topics. Therefore, I hope you could kindly offer your guidance.
Correcting the model errors and refine again in phenix with standard setting.
Hi Guan-Yi,
I can share my strategy of refining any crystal structure solved using the Molecular Replacement Method.
If the search model you used to solve the structure is from your earlier crystal structure, check your notes if there were any problematic regions.
If the model you used to solve structure was from PDB, download the structure with electron density maps, check problematic places
(in Coot -> Validate -> Difference Map Peaks) and create your list.
Usually, I refine the initial solution in REFMAC until the R factors converge to some values. At this point I would check Difference Map Peaks if there are any conformational differences in the structure. Will work on those regions and fix them if that is possible. Do not overdue, do not try to model peaks if you feel it could be wrong (you can build those regions later, when R factors will drop lower and your electron density maps will look better).
At this point I am modeling water molecules using ARP/wARP and refining in REFMAC until R factors converge. I would check Difference Map Peaks and try to fix problems. Some places in the structure could be not well resolved because regions could be disordered. I would not push hard to fix them at this moment. At this point I would check the whole structure in COOT, from the first protein residue till the last water molecule. I would fix side chains, model alternative conformations, add/delete waters based on the electron density peaks. If there are density peaks which do not belong to the protein or water I would try to identify the possible ligands molecules (buffer, salt, ligand) and model them. After the check was complete I would refine the structure in REFMAC until R factors converge. At this point I would generate TLS groups using the web server and create *.tls file and run several round of REFMAC refinement and going back to problematic places if any left and would try to fixi them. If any changes were introduced to the model then I would run several rounds of refinement in REFMAC and then would run MolProbity to check the quality of the model and clashes. Fixing problems from the validation report and preparing your structure for deposition.
If I have enough time I would check the structure one more time and make sure that there are not positive/negative peaks with high sigma levels. Running several cycles of refinement and creating the final model and final electron density maps. The indication of good quality structure are not only R values the quality of electron density maps are important too.
Hope this helps.
Good luck!
- Badges
- Science topic
- Similar topics
- Chemistry
- Nanotechnology
- Nanostructures
- Thin Films