Hi,

I would like to seek some advices on experimental phasing.

I have datasets of the following:
1. Around 4-6 datasets per crystal collected at sulfur edge. There's no 
anomalous signal but I used one of the best dataset collected as the native 
dataset (nat2).
2. Peak (aupeak) and inflexion (auinf) for a gold derivative
3. Peak for a hg derivative (hgE3pk)
4. One native dataset

 Here's the scaleit summary for the different datasets

 Table: The Rfactor

 Riso           |F_nat2      FP_hgE3pk   F_aupeak    F_auinf
 ----------------------------------------------------------------
 F_nat2         |            0.380       0.239       0.239
 FP_hgE3pk      |0.380                   0.388       0.379
 F_aupeak       |0.239       0.388                   0.126
 F_auinf        |0.239       0.379       0.126


 Table: Normal Probability for acentric data


 Normal Prob.   |F_nat2      FP_hgE3pk   F_aupeak    F_auinf
 ----------------------------------------------------------------
 F_nat2         |            3.112       4.661       5.057
 FP_hgE3pk      |3.112                   2.803       2.777
 F_aupeak       |4.661       2.803                   2.120
 F_auinf        |5.057       2.777       2.120


 Table: Normal Probability for Centric data


 Normal Prob.   |F_nat2      FP_hgE3pk   F_aupeak    F_auinf
 ----------------------------------------------------------------
 F_nat2         |            2.688       3.127       3.091
 FP_hgE3pk      |2.688                   1.966       2.017
 F_aupeak       |3.127       1.966                   1.288
 F_auinf        |3.091       2.017       1.288


Table: Anomalous Differences ( FPHi+ v. FPHi-)


 Anom difference          |nref_cent   Prob_cent   nref_acent  Prob_acent  
Rfactor
 
--------------------------------------------------------------------------------------
 F(+)_hgE3pk v F(-)_hgE3pk|361         0.346       1538        0.732       0.209
 F_aupeak(+) v F_aupeak(-)|353         0.020       1552        1.094       0.104
 F_auinf(+) v F_auinf(-)  |557         0.024       2949        1.200       0.105


Sulfur SAD with the datasets collected at the sulfur edges was not successful.
- There are 14 S in the 309 amino acids long protein.
- These datasets range from 2.5A to 3A.
- No anomalous signal can be detected even (all the CCanom << 0.1 across all 
the resolution shells)

SAD with the Hg peak datasets wasn't successful too
- Unable to find the positions of the Hg atoms with shelxd/phenix hyss

But!

I was able to find 3 au atoms using shelxD using the SIRAS (SAD/MAD did not 
work) protocol with aupeak (resolution to 4A) and nat2 (resolution to 2.5A).
- shelxE built about 2/3 of the model with nice (but disconnected) helices.
- After density modification, the map improved with obvious solvent edges. 
However, after rounds of refinement using refmac, the Rfree remains >0.5.
- Autobuilding using arp/warp and phenix autobuild did not complete the model. 
In fact, only 1/3 of protein was build as loops.
- I attempted to build the rest of the model and connect the helices manually 
using coot but there's a lot of ambiguous electron densities extending in two 
different directions.
I can't really comprehend it. It is so near yet so far.
Is the structure really solved? If it's solved, is there anything that I could 
do to build the model? Can I make use of the other datasets that I have to help 
improve the phases?

Really hope to hear from you soon! Many thanks in advance!!

Cheers,
Kelly


Reply via email to