Hi Pavel Just to let you know, I ran phenix.refine with and without weight optimization and I summarize the results below (R, clashscore, Ramachandran and rotamer outliers) without - 27.9/32.1, 21.9, 4.71, 11.08 with - 32.8/35.1, 8.6, 3.89, 6.44 without (torsion-angle NCS) - 27.9/31.0, 26.5, 4.1, 10.82 So, it seems weight optimization is trying to lower the clashes and outliers at the expense of R factor. On Wed, May 20, 2015 at 11:39 AM, Pavel Afonine wrote:

Hi Mohamed,

I am refining a low resolution structure (3.8 A). After I fix most of the

...

outliers (red bars) in Coot, phenix.refine is causing my structure to be worse than the start with 13 % Ramachandran outlier and 13 % rotamer outlier.

could you please send me data and model (before refinement) files then I will have a look. Refinement at low resolution is generally tricky and often requires using beyond the default settings such as secondary structure, rotamer and Ramachandran plot restraints, NCS (at 3.8A Cartesian NCS or even NCS constraints might be a better option).

If you send me files I will try plausible options to see what can be done.

Pavel

Hi Pavel I have two more questions: 1. If I see a small drop in R/Rfree when I request solvent update (I over-rode the default cut-off of 2.8 A), should I keep those waters? 2. Where can I find the values for favored phi/psi values for Ramachandran? Using the FEM map, I can see some residues flagged as outliers but when they are fixed in Coot (real space refine tool) to a favored region, they actually come out of the density. This makes me suspect that those outliers are real. When I use Ramachandran restraints, obviously the outliers are 'fixed' but clashscore increases. In other cases, the tweak needed to fix them are quite small. 3. For publication, do I need to make an image of each outlier that was not fixed? Thanks. On Thu, May 21, 2015 at 11:32 PM, Tristan Croll wrote:

...

So, it seems weight optimization is trying to lower the clashes and outliers at the expense of R factor.

That's certainly not a bad thing. Remember, the object of refinement is to get the best possible model, not the best possible R-factor. I suspect that what these results are trying to tell you is that there's something fundamentally wrong somewhere in your model - something out of register, perhaps? Some close inspection of regions where lots of outliers appear may be in order.

Tristan Croll Lecturer Faculty of Health School of Biomedical Sciences Institute of Health and Biomedical Engineering Queensland University of Technology 60 Musk Ave Kelvin Grove QLD 4059 Australia +61 7 3138 6443

This email and its attachments (if any) contain confidential information intended for use by the addressee and may be privileged. We do not waive any confidentiality, privilege or copyright associated with the email or the attachments. If you are not the intended addressee, you must not use, transmit, disclose or copy the email or any attachments. If you receive this email by mistake, please notify the sender immediately and delete the original email.

On 22 May 2015, at 8:27 am, mohamed noor wrote:

Hi Pavel

Just to let you know, I ran phenix.refine with and without weight optimization and I summarize the results below (R, clashscore, Ramachandran and rotamer outliers)

without - 27.9/32.1, 21.9, 4.71, 11.08 with - 32.8/35.1, 8.6, 3.89, 6.44 without (torsion-angle NCS) - 27.9/31.0, 26.5, 4.1, 10.82

So, it seems weight optimization is trying to lower the clashes and outliers at the expense of R factor.

On Wed, May 20, 2015 at 11:39 AM, Pavel Afonine wrote:

...

Hi Mohamed,

I am refining a low resolution structure (3.8 A). After I fix most of

...

the outliers (red bars) in Coot, phenix.refine is causing my structure to be worse than the start with 13 % Ramachandran outlier and 13 % rotamer outlier.

could you please send me data and model (before refinement) files then I will have a look. Refinement at low resolution is generally tricky and often requires using beyond the default settings such as secondary structure, rotamer and Ramachandran plot restraints, NCS (at 3.8A Cartesian NCS or even NCS constraints might be a better option).

If you send me files I will try plausible options to see what can be done.

Pavel

_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]