help me after several refmac

From: 王瑞 <wangrui704
Date: 2011/10/13

hello everyone:

    Excuse me, could anyone give me some suggestions? Afetr several cycles of refmac, it give me such a result:               

                                       Initial    Final

                    R factor    0.2540   0.2531

                       R free    0.3499   0.3500

     Rms BondLength    0.0151   0.0151

      Rms BondAngle    2.0266   2.0393

     Rms ChirVolume    0.1179   0.1188

 In fact, I have tried model building & continue refmac, but the R free is still above 0.34. The resolution is  43.728 to  2.088Å and space group is C 2 2 21 . What I should to do? 

Thanks

----------
From: Roger Rowlett
Possibility of twinning? A simple way to check this at this point is to turn on the twin option in refmac and examine the log file. Twinning can cause you to get stuck at high R values in unaccounted for.
Roger Rowlett

----------
From: Tim Gruene

Dear ,

- - how complete is your model compared to the sequence expected in the
crystal? If it is not very complete, yet, your R-values might diverge

- - Your RMS BondAngle is pretty high. Set the matrix weight manually to a
lower number (0.1 or less) until the BondAngle is around 1.6

- - try arp/warp or any other autobuilding tool. If your data are fine,
they should produce a decent model.

Having said that you might also go back and check your data processing:
did you cut the resolution appropriately? Otherwise you may include a
lot a noise during refinement which will destabilise the refinement.

Tim

- --
- --
Dr Tim Gruene
Institut fuer anorganische Chemie
Tammannstr. 4
D-37077 Goettingen





----------
From: Eleanor Dodson

 In fact, I have tried model building&  continue refmac, but the R free is

Hard to say without seeing your map, model is probably correct.. Have you built waters? Are there errors? How good is your data?

 Eleanor

----------
From: Yuri Pompeu
Echoing whats been said:
1- Are you sure your crystal really is in C 2 2 21? If so How good is your data (completeness, Rmerge, etc...)
2-Could have twinning? I recently just got done working on a structure that could be scaled in C 2 2 21 but turned out to really be an almost perfect P21 twin.
(of course in monoclinic there are certain conditions for twinning...)
3- How good is your model, is it complete, all the waters? Missing protein or DNA?
 HTH

Yuri

----------
From: Pete Meyer
Similar to Tim's suggestion, but your low resolution limit may be too low (check refmac's chart of R vs resolution to confirm this).

Pete

----------
From: Xiaopeng Hu
Dear Yuri,
Could you give out the details? Such as how you found the twins?

Xiaopeng

----------
From: 王瑞 <wangrui704

After L-test,the result are follows:

TWINNING ANALYSIS:

First principles calculation of potential twinning operators using code by Andrey Lebedev:

First principles calculation has found 0 potential twinning operators

No twinning detected

<!--SUMMARY_END--></FONT></B>

$TABLE: L test for twinning:

$GRAPHS: cumulative distribution function for |L|:0|1x0|1:1,2,3,4:

$$ |L| Observed Expected_untwinned Expected_twinned $$

$$

0.000000 0.000000 0.000000 0.000000

0.050000 0.054127 0.050000 0.074938

0.100000 0.101795 0.100000 0.149500

0.150000 0.148706 0.150000 0.223312

0.200000 0.196638 0.200000 0.296000

0.250000 0.244675 0.250000 0.367188

0.300000 0.293074 0.300000 0.436500

0.350000 0.342463 0.350000 0.503563

0.400000 0.390543 0.400000 0.568000

0.450000 0.440510 0.450000 0.629437

0.500000 0.489954 0.500000 0.687500

0.550000 0.541145 0.550000 0.741812

0.600000 0.592157 0.600000 0.792000

0.650000 0.643445 0.650000 0.837688

0.700000 0.695155 0.700000 0.878500

0.750000 0.747505 0.750000 0.914062

0.800000 0.802393 0.800000 0.944000

0.850000 0.856546 0.850000 0.967938

0.900000 0.909764 0.900000 0.985500

0.950000 0.961975 0.950000 0.996313

1.000000 1.000000 1.000000 1.000000

$$

在 2011年10月13日 下午11:42，Xiaopeng Hu

----------
From: Eleanor Dodson
This Ltest certainly doesnt look twinned.

But your R factors are indictive of "mostly correct but still problems".

Didnt someone suggest running Arp-Warp to rebuild the structure?
This might reinterpret any dodgy bits.

Eleanor