Re: [ccp4bb] High Rfree: Phasing issue or partial crystal disorder

[Andreas Förster]

Dear Pravin,

this brings me to re-emphasize a point I've just made at RapiData.  If
you're working with PILATUS or EIGER detectors, there is no good reason not
to spread your dose over at least 360° of data.  You will get better
statistics than if you collect 57.2° or whatever your strategy software
recommends - and you can laugh P1 in the face if it comes your way.  Just
make sure to adjust expose time or attenuation accordingly when expanding
your data collection to 360°.

This is of no help after data collection, but since you collected data for
SeMet SAD, you probably already have 360° or more and can thus work in P1
(as James suggested) without much pain.

All best.


Andreas



On Thu, Apr 20, 2017 at 12:11 AM, James Holton 
wrote:

>
> In situations like this I always try dropping to P1.  Even if the data are
> highly incomplete in P1 you can still refine it.  Difference maps are
> degraded by poor completeness, but you still might see something.  But
> either way, the R-factors will tell you something.  if dropping to P1
> solves your R-factor problems then you know you were in the wrong space
> group.
>
> The biggest caveat to dropping symmetry operators (and essentially
> replacing them with NCS operators) is that you want to be VERY careful not
> to mix your working and free sets.  The best way to do this is pick your
> free set with the highest-possible symmetry given your lattice, and then
> expand that to P1 using the CCP4 program "CAD".  Then you change the space
> group in "CAD" and it will chunk out the right asymmetric unit.  Oh, and
> then use PDBSET to expand your model to P1 as well.
> All this is done automatically by the program Zanuda, but even Zanuda
> cannot un-merge data.  You need to provide the P1 structure factors and
> then see what it tells you.
>
> Does that help?
>
> -James Holton
> MAD Scientist
>
> On 4/18/2017 5:56 PM, gnufreebsd wrote:
>
> Dear Pravin
> for a kinase, n lobe is quite flexible ,  especially beta1 and beta2, and
> residues beyond theses two strands.
>
>
> best regards
> tiger
>
> 发自 WPS邮箱客戶端 
> 在 Pravinkumar Jagtap  
> ，2017年4月12日 上午2:55写道：
>
> Dear All,
> I am stuck with this problem for 2 months and hope you could help.
>
> We have a 2.1 A dataset for a 380 amino acid long protein. The space group
> is I4 (single molecule in asymmetric unit, 48% solvent content) and the
> dataset is quite perfect (no obvious pathologies). The protein itself is
> organised in 2 lobes (N and C terminal lobes). The sequence identity to
> nearest homologue structure is 17%.
>
> We could  get the phases by SeMet SAD phasing (3A resolution dataset, 5
> SeMet (excluding N-terminal Met), 3 full occupancy SeMet in C-terminal lobe
> and 2 partial occupancy (~0.5 each; present on surface) SeMet in N-terminal
> lobe). Automated  model building (at 2.1 A) yielded nice model for the
> C-terminal lobe (215 residues)  and manually I could build parts (around 80
> residues) of N-terminal lobe with high confidence. In addition we could
> also build a ligand which is sandwiched between C and N terminal lobe.
>
> However the Rfree is stuck at 0.39 (Rwork 0.33). There is indeed some
> patchy density left at the N terminal lobe but as it is discontinuous, I
> cannot build anything in it (except lots of water molecules). In total I am
> missing around 85 residues. These residues are predicted to be present in
> secondary structure (and not flexible).
>
> As I have around 75-80% model built, I would expect that I would have all
> the phases and  should get nice density for the remaining part. But as I
> dont see it, could the rest part be flexible? But again, this is not
> reflected in the R factors (I would then expect low Rfree).
>
> Could it be that I still lack phases (due to partial occupancy of SeMeth
> in N-terminal lobe ) and have to try to get them by heavy metal soaking, or
> there is disorder in the N-terminal lobe? I have also tried solving
> different datasets for same crystal but this has not been useful.
>
> Regards,
> Pravin.
>
>
>


-- 

Andreas Förster, Ph.D.
MX Application Scientist, Scientific Sales
Phone: +41 56 500 2100 <+41%2056%20500%2021%2000> | Direct: +41 56 500 21 76
<+41%2056%20500%2021%2076> | Email: andreas.foers...@dectris.com
DECTRIS Ltd. | Taefernweg 1 | 5405 Baden-Daettwil | Switzerland |
www.dectris.com

[image: LinkedIn] 
 [image: facebook]







Confidentiality Note: This message is intended only for the use of the
named
recipient(s) and may contain confidential and/or privileged information. If
you
are not the intended recipient, please contact the sender and delete the
message.
Any unauthorized use of the information contained in this message is
prohibited.

Re: [ccp4bb] High Rfree: Phasing issue or partial crystal disorder

[James Holton]

In situations like this I always try dropping to P1.  Even if the data 
are highly incomplete in P1 you can still refine it. Difference maps are 
degraded by poor completeness, but you still might see something.  But 
either way, the R-factors will tell you something.  if dropping to P1 
solves your R-factor problems then you know you were in the wrong space 
group.


The biggest caveat to dropping symmetry operators (and essentially 
replacing them with NCS operators) is that you want to be VERY careful 
not to mix your working and free sets.  The best way to do this is pick 
your free set with the highest-possible symmetry given your lattice, and 
then expand that to P1 using the CCP4 program "CAD".  Then you change 
the space group in "CAD" and it will chunk out the right asymmetric 
unit.  Oh, and then use PDBSET to expand your model to P1 as well.


All this is done automatically by the program Zanuda, but even Zanuda 
cannot un-merge data.  You need to provide the P1 structure factors and 
then see what it tells you.


Does that help?

-James Holton
MAD Scientist

On 4/18/2017 5:56 PM, gnufreebsd wrote:

Dear Pravin
for a kinase, n lobe is quite flexible , especially beta1 and beta2, 
and residues beyond theses two strands.



best regards
tiger

发自 WPS邮箱客戶端 
在 Pravinkumar Jagtap ，2017年4月12日 
上午2:55写道：


Dear All,
I am stuck with this problem for 2 months and hope you could help.

We have a 2.1 A dataset for a 380 amino acid long protein. The
space group is I4 (single molecule in asymmetric unit, 48% solvent
content) and the dataset is quite perfect (no obvious
pathologies). The protein itself is organised in 2 lobes (N and C
terminal lobes). The sequence identity to nearest homologue
structure is 17%.

We could  get the phases by SeMet SAD phasing (3A resolution
dataset, 5 SeMet (excluding N-terminal Met), 3 full occupancy
SeMet in C-terminal lobe and 2 partial occupancy (~0.5 each;
present on surface) SeMet in N-terminal lobe). Automated  model
building (at 2.1 A) yielded nice model for the C-terminal lobe
(215 residues)  and manually I could build parts (around 80
residues) of N-terminal lobe with high confidence. In addition we
could also build a ligand which is sandwiched between C and N
terminal lobe.

However the Rfree is stuck at 0.39 (Rwork 0.33). There is indeed
some patchy density left at the N terminal lobe but as it is
discontinuous, I cannot build anything in it (except lots of water
molecules). In total I am missing around 85 residues. These
residues are predicted to be present in secondary structure (and
not flexible).

As I have around 75-80% model built, I would expect that I would
have all the phases and  should get nice density for the remaining
part. But as I dont see it, could the rest part be flexible? But
again, this is not reflected in the R factors (I would then expect
low Rfree).

Could it be that I still lack phases (due to partial occupancy of
SeMeth in N-terminal lobe ) and have to try to get them by heavy
metal soaking, or there is disorder in the N-terminal lobe? I have
also tried solving different datasets for same crystal but this
has not been useful.

Regards,
Pravin.

Re: [ccp4bb] High Rfree: Phasing issue or partial crystal disorder

[Eleanor Dodson]

HOW sure are you of the spacegroup? The only difference between I4 ans I41
absences is that l=2n is absent for I4, l=4n for I 41.

If you have the wrong choice half your symmetry equiv molecules will be
corret but not the others..

Getting the screw axis wrong is a good way to get a reasonable but not
acceptable R factor

Eleanor

On 12 April 2017 at 16:26, Phoebe A. Rice <pr...@uchicago.edu> wrote:

> Hi,
>   Sometimes automated model building needs more manual intervention than
> one might expect, although it sounds like you've already carefully
> inspected the "good" regions.
>   Could you use a model of the N-ter (from a homolog) simply to create a
> solvent envelope, then see if solvent flattening (or more sophisticated
> modern solvent-massaging tricks) improves the density in that region?
> Phoebe
>
> --
> *From:* CCP4 bulletin board [CCP4BB@JISCMAIL.AC.UK] on behalf of
> Pravinkumar Jagtap [pravinja...@gmail.com]
> *Sent:* Tuesday, April 11, 2017 1:54 PM
> *To:* CCP4BB@JISCMAIL.AC.UK
> *Subject:* [ccp4bb] High Rfree: Phasing issue or partial crystal disorder
>
> Dear All,
> I am stuck with this problem for 2 months and hope you could help.
>
> We have a 2.1 A dataset for a 380 amino acid long protein. The space group
> is I4 (single molecule in asymmetric unit, 48% solvent content) and the
> dataset is quite perfect (no obvious pathologies). The protein itself is
> organised in 2 lobes (N and C terminal lobes). The sequence identity to
> nearest homologue structure is 17%.
>
> We could  get the phases by SeMet SAD phasing (3A resolution dataset, 5
> SeMet (excluding N-terminal Met), 3 full occupancy SeMet in C-terminal lobe
> and 2 partial occupancy (~0.5 each; present on surface) SeMet in N-terminal
> lobe). Automated  model building (at 2.1 A) yielded nice model for the
> C-terminal lobe (215 residues)  and manually I could build parts (around 80
> residues) of N-terminal lobe with high confidence. In addition we could
> also build a ligand which is sandwiched between C and N terminal lobe.
>
> However the Rfree is stuck at 0.39 (Rwork 0.33). There is indeed some
> patchy density left at the N terminal lobe but as it is discontinuous, I
> cannot build anything in it (except lots of water molecules). In total I am
> missing around 85 residues. These residues are predicted to be present in
> secondary structure (and not flexible).
>
> As I have around 75-80% model built, I would expect that I would have all
> the phases and  should get nice density for the remaining part. But as I
> dont see it, could the rest part be flexible? But again, this is not
> reflected in the R factors (I would then expect low Rfree).
>
> Could it be that I still lack phases (due to partial occupancy of SeMeth
> in N-terminal lobe ) and have to try to get them by heavy metal soaking, or
> there is disorder in the N-terminal lobe? I have also tried solving
> different datasets for same crystal but this has not been useful.
>
> Regards,
> Pravin.
>
>

Re: [ccp4bb] High Rfree: Phasing issue or partial crystal disorder

[Phoebe A. Rice]

Hi,
  Sometimes automated model building needs more manual intervention than one 
might expect, although it sounds like you've already carefully inspected the 
"good" regions.
  Could you use a model of the N-ter (from a homolog) simply to create a 
solvent envelope, then see if solvent flattening (or more sophisticated modern 
solvent-massaging tricks) improves the density in that region?
Phoebe


From: CCP4 bulletin board [CCP4BB@JISCMAIL.AC.UK] on behalf of Pravinkumar 
Jagtap [pravinja...@gmail.com]
Sent: Tuesday, April 11, 2017 1:54 PM
To: CCP4BB@JISCMAIL.AC.UK
Subject: [ccp4bb] High Rfree: Phasing issue or partial crystal disorder

Dear All,
I am stuck with this problem for 2 months and hope you could help.

We have a 2.1 A dataset for a 380 amino acid long protein. The space group is 
I4 (single molecule in asymmetric unit, 48% solvent content) and the dataset is 
quite perfect (no obvious pathologies). The protein itself is organised in 2 
lobes (N and C terminal lobes). The sequence identity to nearest homologue 
structure is 17%.

We could  get the phases by SeMet SAD phasing (3A resolution dataset, 5 SeMet 
(excluding N-terminal Met), 3 full occupancy SeMet in C-terminal lobe and 2 
partial occupancy (~0.5 each; present on surface) SeMet in N-terminal lobe). 
Automated  model building (at 2.1 A) yielded nice model for the C-terminal lobe 
(215 residues)  and manually I could build parts (around 80 residues) of 
N-terminal lobe with high confidence. In addition we could also build a ligand 
which is sandwiched between C and N terminal lobe.

However the Rfree is stuck at 0.39 (Rwork 0.33). There is indeed some patchy 
density left at the N terminal lobe but as it is discontinuous, I cannot build 
anything in it (except lots of water molecules). In total I am missing around 
85 residues. These residues are predicted to be present in secondary structure 
(and not flexible).

As I have around 75-80% model built, I would expect that I would have all the 
phases and  should get nice density for the remaining part. But as I dont see 
it, could the rest part be flexible? But again, this is not reflected in the R 
factors (I would then expect low Rfree).

Could it be that I still lack phases (due to partial occupancy of SeMeth in 
N-terminal lobe ) and have to try to get them by heavy metal soaking, or there 
is disorder in the N-terminal lobe? I have also tried solving different 
datasets for same crystal but this has not been useful.

Regards,
Pravin.

Re: [ccp4bb] High Rfree: Phasing issue or partial crystal disorder

[Pravinkumar Jagtap]

Dear Mark,
Thanks for your reply. I have already tried P1 SG (we do have 360 degree
data) and Zanuda server without success. I do now have crystals in
different conditions (and visually they are of different shape). In
parallel I am trying heavy metal soaking and hope one of the strategy will
yield better model. But its good to know that other people had such cases
as there is not much literature about this on the internet.

Thanks and regards,
Pravin.

On Wed, Apr 12, 2017 at 9:56 AM, Mark J van Raaij 
wrote:

> Dear Pravin,
>
> we've had a couple of these unfortunately, in our case the only solution
> has been to find alternative crystal forms. Or several crystal forms in
> which different domains are disordered, but allowing to make a composite,
> complete structure for interpretation.
> Your N-terminal domain is probably is several alternative positions in
> different copies of the protein in the crystals. This would give rise to
> disordered density, but not complete flexibility, and thus highish Rs.
> However, before being sure, you should probably try integrating in lower
> symmetry space-groups, going down to P1 if necessary. If you don't have a
> dataset of 180 or even better 360 degrees perhaps collect it from a fresh
> crystal. And check diffraction images carefully for missed weak spots and a
> possible larger cell.
> In this case, I don't think better phases would help, but you never know.
> I would try placing the nearest structural homologue of the N-terminal
> domain as well as you can and see if some density appears, that way you
> have corrected the solvent envelope more or less and this could maybe help
> a bit.
>
> Greetings,
>
> Mark
>
> Mark J van Raaij
> Dpto de Estructura de Macromoleculas
> Centro Nacional de Biotecnologia - CSIC
> calle Darwin 3
> E-28049 Madrid, Spain
> tel. (+34) 91 585 4616 <+34%20915%2085%2046%2016>
> http://wwwuser.cnb.csic.es/~mjvanraaij
>
> On 11 Apr 2017, at 20:54, Pravinkumar Jagtap 
> wrote:
>
> Dear All,
> I am stuck with this problem for 2 months and hope you could help.
>
> We have a 2.1 A dataset for a 380 amino acid long protein. The space group
> is I4 (single molecule in asymmetric unit, 48% solvent content) and the
> dataset is quite perfect (no obvious pathologies). The protein itself is
> organised in 2 lobes (N and C terminal lobes). The sequence identity to
> nearest homologue structure is 17%.
>
> We could  get the phases by SeMet SAD phasing (3A resolution dataset, 5
> SeMet (excluding N-terminal Met), 3 full occupancy SeMet in C-terminal lobe
> and 2 partial occupancy (~0.5 each; present on surface) SeMet in N-terminal
> lobe). Automated  model building (at 2.1 A) yielded nice model for the
> C-terminal lobe (215 residues)  and manually I could build parts (around 80
> residues) of N-terminal lobe with high confidence. In addition we could
> also build a ligand which is sandwiched between C and N terminal lobe.
>
> However the Rfree is stuck at 0.39 (Rwork 0.33). There is indeed some
> patchy density left at the N terminal lobe but as it is discontinuous, I
> cannot build anything in it (except lots of water molecules). In total I am
> missing around 85 residues. These residues are predicted to be present in
> secondary structure (and not flexible).
>
> As I have around 75-80% model built, I would expect that I would have all
> the phases and  should get nice density for the remaining part. But as I
> dont see it, could the rest part be flexible? But again, this is not
> reflected in the R factors (I would then expect low Rfree).
>
> Could it be that I still lack phases (due to partial occupancy of SeMeth
> in N-terminal lobe ) and have to try to get them by heavy metal soaking, or
> there is disorder in the N-terminal lobe? I have also tried solving
> different datasets for same crystal but this has not been useful.
>
> Regards,
> Pravin.
>
>
>


-- 
Pravin Kumar Jagtap,
Department Chemie
Lehrstuhl für Biomolekulare NMR-Spektroskopie
Technische Universität München
Lichtenbergstr. 4
85747 Garching / Munich
Germany.


Helmholtz Zentrum München
Office 110.
Building 43b
Institute for Structural Biology (STB)
Deutsches Forschungszentrum für Gesundheit und Umwelt (GmbH)
Ingolstädter Landstr. 1
85764 Neuherberg

Re: [ccp4bb] High Rfree: Phasing issue or partial crystal disorder

[Mark J van Raaij]

Dear Pravin,

we've had a couple of these unfortunately, in our case the only solution has 
been to find alternative crystal forms. Or several crystal forms in which 
different domains are disordered, but allowing to make a composite, complete 
structure for interpretation.
Your N-terminal domain is probably is several alternative positions in 
different copies of the protein in the crystals. This would give rise to 
disordered density, but not complete flexibility, and thus highish Rs.
However, before being sure, you should probably try integrating in lower 
symmetry space-groups, going down to P1 if necessary. If you don't have a 
dataset of 180 or even better 360 degrees perhaps collect it from a fresh 
crystal. And check diffraction images carefully for missed weak spots and a 
possible larger cell.
In this case, I don't think better phases would help, but you never know. I 
would try placing the nearest structural homologue of the N-terminal domain as 
well as you can and see if some density appears, that way you have corrected 
the solvent envelope more or less and this could maybe help a bit.

Greetings,

Mark

Mark J van Raaij
Dpto de Estructura de Macromoleculas
Centro Nacional de Biotecnologia - CSIC
calle Darwin 3
E-28049 Madrid, Spain
tel. (+34) 91 585 4616
http://wwwuser.cnb.csic.es/~mjvanraaij

> On 11 Apr 2017, at 20:54, Pravinkumar Jagtap  wrote:
> 
> Dear All,
> I am stuck with this problem for 2 months and hope you could help.
> 
> We have a 2.1 A dataset for a 380 amino acid long protein. The space group is 
> I4 (single molecule in asymmetric unit, 48% solvent content) and the dataset 
> is quite perfect (no obvious pathologies). The protein itself is organised in 
> 2 lobes (N and C terminal lobes). The sequence identity to nearest homologue 
> structure is 17%.
> 
> We could  get the phases by SeMet SAD phasing (3A resolution dataset, 5 SeMet 
> (excluding N-terminal Met), 3 full occupancy SeMet in C-terminal lobe and 2 
> partial occupancy (~0.5 each; present on surface) SeMet in N-terminal lobe). 
> Automated  model building (at 2.1 A) yielded nice model for the C-terminal 
> lobe (215 residues)  and manually I could build parts (around 80 residues) of 
> N-terminal lobe with high confidence. In addition we could also build a 
> ligand which is sandwiched between C and N terminal lobe. 
> 
> However the Rfree is stuck at 0.39 (Rwork 0.33). There is indeed some patchy 
> density left at the N terminal lobe but as it is discontinuous, I cannot 
> build anything in it (except lots of water molecules). In total I am missing 
> around 85 residues. These residues are predicted to be present in secondary 
> structure (and not flexible).
> 
> As I have around 75-80% model built, I would expect that I would have all the 
> phases and  should get nice density for the remaining part. But as I dont see 
> it, could the rest part be flexible? But again, this is not reflected in the 
> R factors (I would then expect low Rfree).
> 
> Could it be that I still lack phases (due to partial occupancy of SeMeth in 
> N-terminal lobe ) and have to try to get them by heavy metal soaking, or 
> there is disorder in the N-terminal lobe? I have also tried solving different 
> datasets for same crystal but this has not been useful.
> 
> Regards,
> Pravin.
> 

[ccp4bb] High Rfree: Phasing issue or partial crystal disorder

[Pravinkumar Jagtap]

Dear All,
I am stuck with this problem for 2 months and hope you could help.

We have a 2.1 A dataset for a 380 amino acid long protein. The space group
is I4 (single molecule in asymmetric unit, 48% solvent content) and the
dataset is quite perfect (no obvious pathologies). The protein itself is
organised in 2 lobes (N and C terminal lobes). The sequence identity to
nearest homologue structure is 17%.

We could  get the phases by SeMet SAD phasing (3A resolution dataset, 5
SeMet (excluding N-terminal Met), 3 full occupancy SeMet in C-terminal lobe
and 2 partial occupancy (~0.5 each; present on surface) SeMet in N-terminal
lobe). Automated  model building (at 2.1 A) yielded nice model for the
C-terminal lobe (215 residues)  and manually I could build parts (around 80
residues) of N-terminal lobe with high confidence. In addition we could
also build a ligand which is sandwiched between C and N terminal lobe.

However the Rfree is stuck at 0.39 (Rwork 0.33). There is indeed some
patchy density left at the N terminal lobe but as it is discontinuous, I
cannot build anything in it (except lots of water molecules). In total I am
missing around 85 residues. These residues are predicted to be present in
secondary structure (and not flexible).

As I have around 75-80% model built, I would expect that I would have all
the phases and  should get nice density for the remaining part. But as I
dont see it, could the rest part be flexible? But again, this is not
reflected in the R factors (I would then expect low Rfree).

Could it be that I still lack phases (due to partial occupancy of SeMeth in
N-terminal lobe ) and have to try to get them by heavy metal soaking, or
there is disorder in the N-terminal lobe? I have also tried solving
different datasets for same crystal but this has not been useful.

Regards,
Pravin.