Dear Ting-Wei,

1) Refmac writes model statistics like RMS bond lengths, angles etc. in the header of the ouput pdb. Open the output pdb in your favorite editor and look for REMARK 3 cards. There you find probably all the information you need.

2) You give very little details, so here I can only make some general remarks. I asume that your problem is that the electron density for A' looks broken?

- check your diffraction images to make sure they are ok. Things to look for are ice rings, bad zones etc. I usually look at every 10th frame.

- did you check for twinning?

- do not asume the space group you got from your data processing package is correct. In molecular replacement you should test all space groups that are compatible with the cell dimensions. E.g. in phaser you can use the keyword SGALTERNATIVE ALL.

- From the Mathews coefficients you give, I would say that 3 molecules in the AU are also possible, so you should look how the electron density looks with fewer molecules.

- I guess that A' and B are both proteins? In that case your crystals could also contain only A' or only B molecules, so you should repeat the molecular replacement with only A' and with only B as search model.

- It never hurts to collect another data set, preferably from a crystal grown under different conditions.

Good luck!

Herman

From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of Ting-Wei Jiang
Sent: Thursday, June 09, 2011 9:46 AM
To: [log in to unmask]
Subject: [ccp4bb] regarding refinement and structure determination by MR

Dear experts,
I got two questions regarding refinement and structure determination by MR.

1.I got a dataset of resolution at 2A refined to Rvalue~20% and Rfree~24%
   I want to know if there is any program in CCP4 could help me to check the refined structure in detail.
   For example,the model statistics in CNS could list some information that infer band,angle violation...etc
   Or I should ask what do I need to check before submitting to PDB.

2.A dataset of resolution collected to 2.2A is from native protein crystal(A) and its complex form(A'+B)
whose PDB code is 2QN5 was already determined by MR(use another protein as search model).I planned
to use A' as search model but the A' looks broken.
This number of molecules in AU was predicted to 4~6 using Mathews coefficient.
N/a    M.C. solvent%
   4    2.99    58.92
   5    2.39    48.65
   6    1.99    38.38
The coordinate or orientation of output pdb(as attached figure) from MR are always different since I changed
parameter,such as different resolution,Multi copy,search mode...etc. even the contrast value of every run
is pretty high(>>3)
I really don't what happened with this dataset and any suggestion would be greatly appreciated.