Hi Nika
You said you refine with phenix, correct? And that refining the phaser solution 
yielded in Rwork of 22% and a gap of 7%. For refining the rebuilt model, you 
had to open a new phenix.refine job, did you provide the native mtz there? Or 
did you maybe refine against the output mtz of the phaser refinement? I'm 
asking because you could have run the first refinent w/o Rfree flags, and upon 
choosing the output mtz, accidentally set the flags anew. That would fit to 
Dale's suggestion?



Dr. Matthias Barone
AG Kuehne, Rational Drug Design
Leibniz-Forschungsinstitut für Molekulare Pharmakologie (FMP)
Robert-Rössle-Strasse 10
13125 Berlin
Germany
Phone: +49 (0)30 94793-284
________________________________
From: CCP4 bulletin board <[email protected]> on behalf of Wim Burmeister 
<[email protected]>
Sent: Monday, November 2, 2020 5:04:34 PM
To: [email protected]
Subject: Re: [ccp4bb] R free rising

...or your dataset  may have a much lower resolution than the one your initial 
model was based upon. The discrepancy between Rfree and Rwork seem to indicate 
this.
Best
Wim
________________________________
De: "Eleanor Dodson" <[email protected]>
À: "CCP4BB" <[email protected]>
Envoyé: Lundi 2 Novembre 2020 12:08:17
Objet: Re: [ccp4bb] R free rising

Yes, as Dale says when the FreeR goes up after minor rebuilding you have 
usually somehow picked up a different FreeR set..
This is almost certainly what causes this to happen - you say

This results in R free slightly lower than R work.

Small changes in a well refined structure dont change r factors very much!
Eleanor

On Mon, 2 Nov 2020 at 10:57, Dale Tronrud 
<[email protected]<mailto:[email protected]>> wrote:
On 11/2/2020 2:26 AM, Nika Žibrat wrote:
>
> Hello,
>
>
>
> I am trying to solve an X-ray structure of a protein of which the
> structure is already known. My aim is to only seek for ligands (soaking)
> and interpret any conformational changes. Since I am using a model with
> 100% sequence identity from PDB I am not doing Autobuild after Molecular
> phasing and continue directly with phenix.refine according to
> reccomendations (10 rounds). In accordance with X-triage I am also using
> NCS default settings in the refinement.
>
>
>
> This refinement produces solid R free and R work values around 0.29 and
> 0.22. The problem becomes when I want to manually edit the structure,
> correct the loops which are changed upon binding of the ligand, and
> correct any outliers. This results in R free slightly lower than R work.
> Upon refining, R work drops normally while R free rises significantly
> (for 0.2 -0.3). I have been trying to crack this for a few days with no
> success.
>
>
>
> I read that slightly lower R free can be normal in such cases but
> nevertheless both R values should drop, and haven’t found anything about
> the big rise of this value after refinement. It feels like I am missing
> something, since this is my first time solving a structure. Any advice?

   This is not normal behavior at all.  Rwrk and Rfree will be roughly
equal only before you perform any refinement.  The R's you report before
your model building sound quite reasonable.  When you manually change
the model you will likely cause both to increase, but you would have to
perform massive changes to get them to equalize at some larger value.

   The only thing I can think of that would cause this is for your
second refinement to be working with a newly created test set.  It is
possible that somehow you have reset your R free flags?  In an MTZ the
full data set is divided into twenty subsets -- one is the test set
while the other nineteen are the working set.  When you ran Refmac the
second time could you have told it to use a different segment as the
test set?

Dale Tronrud

>
>
>
> Thank you,
>
> Nika
>
>
>
>
> ------------------------------------------------------------------------
>
> To unsubscribe from the CCP4BB list, click the following link:
> https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1
>

########################################################################

To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1

This message was issued to members of 
www.jiscmail.ac.uk/CCP4BB<http://www.jiscmail.ac.uk/CCP4BB>, a mailing list 
hosted by www.jiscmail.ac.uk<http://www.jiscmail.ac.uk>, terms & conditions are 
available at https://www.jiscmail.ac.uk/policyandsecurity/

________________________________

To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1


________________________________

To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1

########################################################################

To unsubscribe from the CCP4BB list, click the following link:
https://www.jiscmail.ac.uk/cgi-bin/WA-JISC.exe?SUBED1=CCP4BB&A=1

This message was issued to members of www.jiscmail.ac.uk/CCP4BB, a mailing list 
hosted by www.jiscmail.ac.uk, terms & conditions are available at 
https://www.jiscmail.ac.uk/policyandsecurity/

Reply via email to