Hi Yanming,
    
>>Story ends with my questions:
>>1,It seems the 3rd copy non-exist or globally disordered, under this circumstance, can I say 'because of the 3rd molecule is globally disordered so 
>>>Rfree and R stay high (2mol/a.u with Rfree 30% R 27%, 1.8A data)?

After finding the first 2mol/au followed by refinement when you got R/Rfree of 27%/30%, do the resulting maps indicate presence of a third molecule? If so, the third molecule has to exist. 

>>2,Anything I should worry about in the process finding out the 3rd?

  You mention that the solution for the third molecule obtained by Molrep and Phaser differ. And you prefer the Phaser solution for the third since it results in lower R/Rfree. However, maybe both these solutions are not totally correct which results in the bad refinement with 3 mol/asu.
Does this third solution match the density obtained in (1) above for the third molecule after refinement with only 2mols/asu? I think it is more likely to be incorrect placement of third molecule rather than global disorder.


>>3,Are there suggestions to further lower down Rfree (the reasonable Rfree should be around 22%)?
Check several different placements of the third molecule. Try simulated annealing refinement. 
What about twinning possibility? You can also use Phaser to try mr in both P43212 and P43 groups just to make sure that the other space group just to make sure your sg choice is correct.

Regards,
Debanu.

--
Debanu Das,
JCSG,
SSRL.


-----Original Message-----
From: CCP4 bulletin board [mailto:[log in to unmask]] On Behalf Of Yanming Zhang
Sent: Wednesday, August 15, 2007 6:25 PM
To: [log in to unmask]
Subject: [ccp4bb] Structure help

Hi ,
Please help me on this structure:
Data 1.8A: cell 84.892 84.892 172.580 90 90 90 with tetrahedral index Space group: P43212 or P43, Final refinement Rfree indicates P43 is better so most likely P43(also the systematic absence indicate P43 is it) Matthew coeffient indicates: P43 3mol/a.u(2.58,52%) 2mol/a.u(3.87, 68%)

MR: Both Molerep and Phaser find out 2mol/au initially. Refinement reaches Rfree 30% , R 27% finally. WE think, comparing with 1.8 resolution, The R and Rfree are too high. Also considering the 1.8A resolution, there should be 3 molecules in total. So we engaged into the effort to find out the 3rd molecule.

Phased MR by MOLREP did find out the 3rd, so did Phaser(fix two, find out the 3rd). Both solutions gave relatively poorer statistics but packing is perfect. The 3rd molecule, solved by phased MR of MOLREP and PHaser respectively, can not superpose  showing slightly different orientations, but position is similar. 
Refinement indicates PHased MOLREP MR gives lower Rfree.

Using totally 3 molecule, When I restrain or constrain NCS during refinement(used TLS),R free goes way up (R~30, Rfree > 36%). If I refine without NCS,R factors slip right away to 27/30 but this strange thing
happens: two copies of the original solution refines very well - low B factors,very good looking map. The 3rd copy, however, B factors are high and the map is bad, no good density match with the model.And the 3rd molecule did no help to Rfree or R.

Story ends with my questions:
1,It seems the 3rd copy non-exist or globally disordered, under this circumstance, can I say 'because of the 3rd molecule is globally disordered so Rfree and R stay high (2mol/a.u with Rfree 30% R 27%, 1.8A data)?
2,Anything I should worry about in the process finding out the 3rd?
3,Are there suggestions to further lower down Rfree (the reasonable Rfree should be around 22%)?

THANKS
Yanming