Just a word of warning -
1) you cant really compare R factors between twinned and untwinned
refinements - they are calculated differently.
2) If the data is twinned the Free R set needs to be selected in the
higher symmetry Laue group and expanded so that twinned pairs are
either both in the Free set or both in the working set.
Eleanor
On 13 October 2013 10:51, Dilip Badjugar <[log in to unmask]> wrote:
> Dear experts,
>
> Yes, you were right the data is twined and shows 1 merohedral twin operator.
> Now I am performing twin refinement, hope the refinement statistics will
> improve. Thank you very much for your valuable suggestions.
>
> Regards
>
> DCB
>
>
>
> On Sun, Oct 13, 2013 at 3:01 PM, Kay Diederichs
> <[log in to unmask]> wrote:
>>
>> Dear Dilip,
>>
>> yes, your data probably are ok, at least based on the value of I/sigma=3.2
>> in the highest shell (and based on that, the data could even be used to 1.6
>> or maybe 1.5A). The overall and the highest shell Rmerge actually do not
>> carry much information. It would be rather interesting to know the Rmerge
>> (or better, Rmeas) in the lowest resolution shell - it should be less than
>> 5% (ideally 2%).
>>
>> You have not answered my question, probably because its meaning was not
>> clear. What I mean is: in space group P3(2)21 and relatives twinning may
>> (and often does) occur. If the twinning fraction is low to medium (0.05 to,
>> say, 0.25), you have no problem with solving the structure and refining it.
>> But R/Rfree may be stuck, because e.g. with a twin fraction of 0.25, your
>> model, even if ideal, can only explain 75% of the diffracted intensity; the
>> remaining 25% will increase the R-factors and lead to unexplainable
>> difference density blobs.
>>
>> The remedy is to analyse the twinning properties with a tool like
>> phenix.xtriage, and to switch on twin refinement in phenix.refine if
>> phenix.xtriage tells you that the data is twinned. The analysis is quite
>> reliable because you can even give your current model to phenix.xtriage .
>>
>> hope that helps,
>>
>> Kay
>>
>> On Sun, 13 Oct 2013 13:07:29 +0530, Dilip Badjugar <[log in to unmask]>
>> wrote:
>>
>> >Dear experts,
>> >
>> >Thank you very much for your valuable the suggestions. Dear Tim, the
>> >electron density map is good and fitted well. Krishnaswamy sir, there are
>> >some unmodelled blobs with positive density, I have added respective
>> > ions
>> >and refined the model using phenix.refine (Rf- 20.56 and Rfree- 25.76).
>> > Now the question, what is the actual resolution of the data? When I
>> > scaled
>> > to 1.7� resolution the data statistics is as follows:
>> >
>> >Rmerge- 25.5 (239)a
>> >
>> > I/sigmaI- 10 (3.2)
>> >
>> >
>> > Completness-100
>> >(100)
>> >
>> > Multiplicity 11.8
>> > (12.1)
>> >
>> >a - indicate the values in the outer shell (1.7-1.79 �)
>> >
>> >The Rmerge at 3.1 � resolution is 11.5 % while at 1.7 � it is 239%. Do I
>> >have enough data to justify 1.7 � resolution? Are those Rfactor values
>> > are
>> >justified with this resolution?Thanking you
>> >
>> >DCB
>> >
>> >
>> >On Sun, Oct 13, 2013 at 12:03 PM, Kay Diederichs <
>> >[log in to unmask]> wrote:
>> >
>> >> Are you certain that the data are not twinned?
>> >>
>> >> HTH
>> >>
>> >> Kay
>> >>
>> >>
>> >
>> >
>> >--
>> >Mr.Dilip C. Badgujar,
>> >Senior Research Fellow,
>> >ACTREC,
>> >Tata Memorial Center,
>> >Sector-22, Kharghar,
>> >Navi Mumbai.
>> >Pin-410210
>> >
>>
>>
>
>
>
> --
> Mr.Dilip C. Badgujar,
> Senior Research Fellow,
> ACTREC,
> Tata Memorial Center,
> Sector-22, Kharghar,
> Navi Mumbai.
> Pin-410210
|