Dear all
I submitted the data and model to PDB_REDO, which returned a model with better statistics (R/Rfree 0.27/0.27 as recalculated with Phenix). However, the geometry was poor with a high number of C-beta deviations. So, I refined again with phenix.refine and asked for automatic water and ion placement with a resulting R/Rfree of 0.22/0.24, about 600 water molecules and 3 Na+ ions.
I guess somehow the model was stuck at a local minimum.
Thanks again for your suggestions (and to PDB_REDO) :)
Mohamed
On Wed, 9 Mar 2016 21:38:41 +0000, Tanner, John J. <[log in to unmask]> wrote:
>Dear Mohamed,
>
>You are right to be concerned that the lack of solvent density could be symptomatic of deeper problems. We had a similar case a few years ago that we could never solve. Here are the details:
>
>Resolution: 2.1 Angstrom
>-Mean((I)/sd(I)) 10.9 34.5 1.7
>-Mn(I) half-set correlation CC(1/2) 0.998 0.998 0.703
>Apparent SG: 82.199 82.199 129.232 90.00 90.00 120.00 P 31 2 1
>VMatt: 2.29
>Solvent: 46%
>Phasing: MR with 60% identical search model
>ASU: 500 residues
>phenix.xtriage:
>-No twinning is suspected.
>-Multivariate Z score L-test: 0.657
>-No significant pseudotranslation is detected.
>
>Refinement using an almost complete protein model gave:
>
>REMARK Final: r_work = 0.3258 r_free = 0.3974 bonds = 0.010 angles = 1.185
>
>The map shows good density for bound nucleotide ligands, but very little solvent density. Refinement with default ordered water building in phenix added only 63 water molecules with negligible improvement in the R-factors or map quality:
>
>REMARK Final: r_work = 0.3145 r_free = 0.3842 bonds = 0.010 angles = 1.133
>
>Refinement with twin operators and/or lower symmetry SGs did not help.
>
>In summary, your concerns are justified.
>
>Good Luck,
>
>Jack Tanner
>
>
>On Mar 8, 2016, at 12:20 AM, Mohamed Noor wrote:
>
>> Dear all
>>
>> I am working on a 2-A structure with an octameric arrangement (363 residues per chain). There are only about 20 water molecules in total, which does not correspond to the rule-of-thumb of one water per residue (Carugo and Bordo 1999). Is this something to worry about?
>>
>> The R/Rfree is currently at around 30/32 %. The protein chains are almost complete except for some difficult loops in some chains.
>>
>> Thanks.
>> Mohamed
|