Re: [ccp4bb] Search by name on the PDB fails

[Gert Vriend]

The MRS server at the CMBI (http://mrs.cmbi.umcn.nl/) is very quick. It 
is not a dedicated ligand searcher, but it finds 23 files with 
tetraethylene glycol. It is just a Google-like text-search engine, but 
much smarter than grep; for example "tetraetylene glycol" suggests that 
you might want to use it with the h... It does this in 0.3 second, so, 
if it doesn't solve your problem, it certainly won't waste much of your 
time.


MRS is open source, so you can install your in-house version (see 
github). But we will keep the cmbi site up and running for a while...


Gert


On 2-10-2018 7:17, Murpholino Peligro wrote:

Hi all.
I was searching for ligands/molecules at 
www.rcsb.org/pdb/ligand/chemAdvSearch.do 
 (middle tab in 
blue). If I try to search with the third option, i.e. "name" of the 
ligand/molecule, it never works. I tried tetraethylene glycol, nitric 
oxide, cisplatin and even the example provided biotin. Fortunately 
specifying the identifier (i.e. 3 letters..2 in some cases)... works.

The problem is...what if I do not know the identifier?

I was wondering if you know another database that might work better 
for me? (or you, if your brain works better with chemical names 
instead of identifiers)



Ps If there is some sort of coordinate file, preferably pdb or xyz, to 
download that will be better.



* Pubchem works with sdf files which I do not like because if I open 
the sdf file for biotin with pymol and then save it as pdb, and then 
open this on coot it looks strange... (png attached)
(not sure if this is pymol's or coot's fault...but that's another 
thread I guess)

Thanks




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






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

Re: [ccp4bb] Can H-clashes be ignored ?

[Gert Vriend]

On 17-8-2018 18:27, Firdous Tarique wrote:

Hello everyone.

I have a basic question. When a validation report of a coordinate is 
generated we often come across a term known as "Too-Close Contacts".

Essentially evey PDB file has at least a few of those.
First of all can somebody please explain me what is the shortest 
interatomic distance between the two atoms which is permissible ?
There isn't one number for that. When two atoms get close to each other 
the balance of repulsive and attractive forces should get close to zero 
(the balance of all forces on one atom should be zero in the static 
representation of the dynamic structure; but this is probably 
un-achievable). If you do energy minimisation with , for example, 
GROMACS or YASARA, you can get at zero summed forces on all atoms, but 
these are zero in the force field of these MD/EM softwares. And each 
time you use another program (or another force field in the same 
program) you get different coordinates by the time all summed forces per 
atom are zero.
Most force fields know a limited number of atom times and then define a 
few functions that determine the forces between those atom types (see eg 
https://swift.cmbi.umcn.nl/teach/B2/bioinf_6.html for a very low-level 
introduction). One of those forces is the so called Van der Waals' force 
(or when integrated, the Van der Waals' energy). This is a description 
of the force between two atoms (actually atom types) as function of 
their distance.
What we now do in validation is estimate the shortest distance for which 
the repulsive force is 'acceptable'. In WHAT_CHECK we did this by simply 
counting in a couple thousand PDB files which which is the frequency of 
narrow distance bins for short distances and from that determine which 
distance only 1 in 1000 pairs should have. From that we subtract a 
safety margin of 0.25-0.4 Angstrom), and any pair of atoms shorter than 
that is flagged. Obviously, this is not optimal, but today still the 
best we think we can do.
H-atoms are an additional problem as we didn't have many protons 
available when WHAT_CHECK was designed. But I assume the people who gave 
you your clash list have done something similar using very high 
resolution structures.
When dealing with H-atoms you need to separate the riding ones, that 
have fixed positions, and the others. The riding ones should worry you 
more, especially in the beginning of the refinement. This because fixing 
these bumps/clashes requires moving non-H atoms. The other H-atoms often 
can be de-clashed by rotating them around the heavy atom they are 
attached to.
 Next, in this list there are Non-H and H columns list, their 
Interatomic distance and Clash overlap. I could see  two types of 
clashes in my validation report. One in which interatomic distance 
between the two atom (one is always a modeled  H) ranges from 
1.7-2.4A, and clash overlap from 04-1.0. The other in which the 
interatomic distance between two atom is greater than 2.2A and the 
clash overlap is between 0.4-0.6 (always between two non H-atoms).
I would clearly start addressing the non-H atoms first. With some luck, 
de-clashing the non-H atoms will solve the H clashes too.


So my question is that out of so many clashes shown in the list which 
are one which actually need to be fixed and can't be ignored specially 
because one of my ligand is an amino acid which is showing lots of 
these H clashes (interatomic distance less than 1.5A).
If you have a lot of such very bad clashes, you might want to reconsider 
the whole refinement, or perhaps even the steps before you started 
refining. One suggestion is to see what PDB_REDO can do for you.


Greetings
Gert

(Ps, feel free to send me your coordinates and I will give you my 
opinion of-the-list).


Regards

Kahkashan





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






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

[ccp4bb] Normalization of B-factors

[Gert Vriend]

TLS groups are a good way to reduce the R-factor without doing very 
stupid things. They give a better fit of coordinates to the data, so, 
when done well, you get a better description of reality without 
(hopefully) over-fitting. Most times, I think, TLS groups are defined by 
the experimentalist and it is rather easy to have some residues in a TLS 
group that perhaps better should not have been in that group. In 
practice this seems not to be a big problem, but when you are going to 
use TLS parameters to compare mobility reduction it seems worth thinking 
about these problems.


That's why we made the BDB (http://www.cmbi.umcn.nl/bdb/), a database in 
which we (for about 80% of the whole PDB) calculated isotropic B-factors 
for all atoms. Those isotropic B-factors might also not be the best 
thing, but when studying mobility aspects of proteins they will work 
better than either looking at TLS values or looking as residual B-factors.


Obviously, the BDB holds data only for deposited PDB files, but if 
anybody contacts me outside the list we might be able to help.


Gert



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

Re: [ccp4bb] Off-topic: protein 2d diagrams

[Gert Vriend]

 Batch mode generation of residue-based diagrams of proteins

Fabien CampagneEmmanuel BettlerGert VriendHarel Weinstein
/Bioinformatics/, Volume 19, Issue 14, 22 September 2003, Pages 
1854–1855,https://doi.org/10.1093/bioinformatics/btg236

Published:
22 September 2003

On 4-8-2018 10:38, Joana Pereira wrote:

Dear CCP4 community,

I have seen many papers with detailed protein 2d diagrams like this one:

However, I can never find a reference for any tool that can compute 
such diagrams from a 3D structure. Do you know of any?


Many thanks and enjoy the warm weather!

Joana


—
Dr. Joana Pereira
Postdoctoral Researcher
Department of Protein Evolution

Max Planck Institute for Developmental Biology
Max-Planck-Ring 5
72076 Tübingen
GERMANY



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






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

Re: [ccp4bb] salt bridges etc..

[Gert Vriend]

Dear Eleanor,

Salt bridges are a compromise between entropy and enthalpy. If, say, an 
Asp and an Arg side chain are a bit restricted in their freedom and the 
charges are close, enthalpy wins, and if they are very exposed, and not 
close at all, entropy wins. The enthalpic gain upon protein folding from 
obtaining one salt bridge has been given many values in the literature, 
but in practice boils down to about 1 kCal/Mole. The enthalpic 
contribution is a bit higher than 1kcal/Mole when the positive and 
negative charges are very close to each other (in which case you loose 
entropy upon folding). Most salt bridges are at the surface where they 
continuously compromise between entropy and enthalpy. So, they move 
around, but most of the time the charges are close together, and that is 
why you can see them with Xray. When the two charged groups come close 
to each other there are always certain local conformations that are 
preferred over others. Those (sometimes multiple) locally preferred 
conformations we see in Xray if we have good crystals. It does not 
matter, however, how many local conformations we observe. It is just one 
salt bridge, and its energetic contribution to protein folding remains 
(very roughly, and this is practical experience for which no good theory 
exists) about 1kCal/Mole.


Gert


On 11-7-2018 16:52, Eleanor Dodson wrote:
How do people decide on what is a salt bridge within a molecule and 
how to count them for those Tables?


I have been looking at 2z2f - paper claims some score..-

But there are several residues in alternate conformation

with NZ A  to OE1A    and NZ A to OE1B  and NZ B to OE1B etc

Is that one salt bridge   or 3 salt bridges

PISA lists salt bridges between molecules but not within a molecule I 
dont think?


Suggestions gratefully received.
Eleanor







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






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

Re: [ccp4bb] validating a homlology model

[Gert Vriend]

Careina,

Feel free to contact me about these things without using the CCP4BB list.

I am sending this to the list with as purpose that others might in the 
future refer people with questions about homology modelling to me. After 
all, although the possibility to build homology models is one of the 
important reasons for having structures deposited in the PDB, it is not 
at all related to crystallography in any way other than that both 
techniques try to get the most precise and accurate coordinates of 
(macro) molecules.


Gert



On 2-3-2018 12:44, Careina Edgooms wrote:

Dear all

What programs are best used for validate homology models? I know of 
molprobity but if there are no coordinates I cannot use it. Is there a 
way to use such programs with homology models?


Also I wish to use pdbepisa for to charaterise dimer interface but 
again for homology model this cannot be done as there is no PDB model. 
Does anybody know way to use PISA software on my own model that is not 
deposited in PDB?


Thank you in advance
Careina

Re: [ccp4bb] Electrostatic Potential: Poisson-Boltzmann criteria

[Gert Vriend]

That is a simple question with a long and tedious answer.

The question is like: "I have this protein crystal, can you suggest some 
ways to solve the structure"?


The best is to leave this to an expert in the field (I am not a real 
expert in this field, but I can give you some pointers of the list, if 
you want)


Greetings

Gert


On 16-12-2017 23:01, chemocev marker wrote:

Hi
I am just calculating the Electrostatic Potential, and I wanted to 
know your opinion which force field is better.What is criteria to 
choose and not to choose the pKA...


best

Jiri

Re: [ccp4bb] secondary structure prediction

[Gert Vriend]

I would try your local bioinformatician (or a remote one if there isn't
any one local).

With the sequence in the mail, I could have given it a shot...

Gert


On 12/6/2017 3:14 PM, zheng zhou wrote:

Dear CCP4 community,

Sorry for the off-topic question. I am trying to design constructs for
structure studies. It only has a homolog structure in PDB with
sequence identity ~20%. When I blast against PDB sequence, there are
quite a few motif hits (30~40aa, identity 40~50%). Any prediction
tools utilize this information?

Thanks for your advice in advance.

Best,

Zheng


Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

[ccp4bb] Developers, users

[Gert Vriend]

In 1996 we wrote a short note on WHAT_CHECK. The fact that protein
structures contain errors caught the community by surprise at that time.
A few weeks later Nature published another note, now by some prominent
crystallographers who stated that WHAT_CHECK produced many false
positive error messages. We were caught by surprise by this note in
Nature. By the way, two of the three authors of that note have by now
apologized for it.

The point that the authors went (very) public with what they believed to
be an error in WHAT_CHECK (but which was later convincingly proven to be
an error in their data and that data was corrected in the PDB a few
years later) was not only very unpleasant for us who wrote WHAT_CHECK,
but also very contra-productive. WHAT_CHECK development was delayed by
many years; partly because a big grant for its development could not be
renewed.

So, as a developer who really suffered from not getting user feedback, I
can tell from first hand that user feedback is not only appreciated, it
also is the honest and scientifically most productive way of dealing
with perceived software problems, and it might even avoid that the
complaining user needs to apologize.

And if a disclaimer needs to be written, then this is obviously best
done by the author of the software for which the disclaimer holds.

Greetings

Gert

Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

[ccp4bb] Regarding Patents

[Gert Vriend]

A related question. If you have a crystal structure and found a novel 
ligand binding site that can be used to regulate protein activity, 
could you patent such "binding site"? If not, how to make the best use 
of such findings?


I would say that the best one can do with important novel 
data/information/knowledge/insights is to publish it so the world can 
benefit from it.


Gert

Re: [ccp4bb] How to deal with the bad omega angles?

[Gert Vriend]

Look at http://swift.cmbi.ru.nl/servers/html/index.html under "structure 
validation" you will find a server that predicts which peptide planes 
most likely need to be flipped. This server is the implementation of:


Detection of trans-cis flips and peptide-plane flips in protein 
structures. 


*Touw*WG,*Joosten*RP,*Vriend*G.

Acta Crystallogr D Biol Crystallogr. 2015 Aug;71(Pt 8):1604-14. doi: 
10.1107/S1399004715008263. Epub 2015 Jul 28.



Good luck,

Gert


On 10-10-2017 15:28, Yang Shi wrote:

Hi, Tristan Croll,

Thanks for your suggestion.I will consider to rebuild these first.

Yang


在 2017年10月10日，下午9:18，Tristan Croll  写道：

Relying on refinement to fix cis peptide bonds for you is unlikely to end well. 
It looks to me like you really need to spend some time investigating and 
manually rebuilding these first.

On 2017-10-10 13:52, 师扬 wrote:

Dear all,
I am refining a model based on a 4.3A EM density map,and there are
some cis-peptides in the beginning model.
By using phenix.real_space_refine with a very low cis-peptide
threshold (0), all the cis-peptide become to the twisted.
The start Omega angle:
cis-proline: 31.63 %
twisted proline: 0.00 %
cis-general: 11.11 %
twisted-general: 0.05 %
The final Omega angle:
cis-proline: 0.00 %
twisted proline: 27.55 %
cis-general: 0.00 %
twisted-general: 6.04 %
My questions are:
1) What is the twisted peptide?
2) Is the amount acceptable at the current resolution?
2) How to refine it?
Thanks in advance!
Yang Shi
【网易自营|30天无忧退货】仅售同款价1/4！MUJI制造商“2017秋冬舒适家居拖鞋系列”限时仅34.9元>>
[1]
Links:
--
[1] http://you.163.com/item/detail?id=1165011&from=web_gg_mail_jiaobiao_9

Re: [ccp4bb] Questions about antibody in crystallization

[Gert Vriend]

The fact that the PDB holds hundreds of FABs and a handful whole ABs 
suggests to me that the latter are hard to get crystals for. So, all the 
more reason for us bioinformaticians that you, experimentalists try it :-)



Gert


On 22-6-2017 20:39, Cheng Zhang wrote:

Hi all,

I have a naive question about antibodies. Many people used Fab 
fragments in crystallization. I am wondering if it is possible to use 
the whole IgG molecule with Fc fragment as well. Or it would be too 
flexible and bad for crystallization?


Thanks,

Cheng


--
-
Cheng Zhang

[ccp4bb] Fwd: Re: [ccp4bb] Structure comparison

[Gert Vriend]

Rather then now start mentioning yet 45 other 3D superposition
softwares, I think a pointer to the Wikipedia should suffice:

https://en.wikipedia.org/wiki/Structural_alignment

The Russian Doll effect (if you align more residues the comparison stats
get worse) is best explained (my opinion) by the CATH people (Orengo
group), and Arthur Lest is the only one (still, I think) who has been
thinking extensively about the cutoff question.

Greetings
Gert


From: Reza Khayat
Sent: Sunday, April 9, 2017 6:07 PM
To: CCP4 bulletin board
Subject: Structure comparison

Hi,

I have refined several structures of a protein from different space groups and 
would like to compare them to one another. Is there a program/software suite 
that would provide an objective comparison of the structures and identify 
regions where the structures are sufficiently different from one another to 
warrant a closer look? I think the most important aspect of the analysis would 
be defining a threshold (possibly based on resolution and structure statistics) 
that would identify sufficient difference between structures. Thanks.

Best wishes,
Reza

Reza Khayat, PhD
Assistant Professor
City College of New York
Department of Chemistry
New York, NY 10031


Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

Re: [ccp4bb] Structure comparison

[Gert Vriend]

Arthur Lesk (at Penn state Univ, I think)m is the only one I know who 
has worked on this topic. I suggest you ask him. The topic you elude to 
is commonly known as the Russian Doll Effect.


If you want to discuss the topic, feel free to Skype me.

Greetings

Gert Vriend


On 10-4-2017 1:37, Reza Khayat wrote:

Hi,

My initial e-mail may have been a bit vague so I'll try to be more specific. 
Superposing the structures and comparing them against one another, while 
appropriate, is a subjective way to do the analysis as I would have to 
subjectively define a threshold that would indicate a difference between the 
structures. My threshold may be grossly different than someone else's 
threshold. I am interested in an objective criterion. One where strong emphasis 
has been put on error analysis and error modeling in terms of both the refined 
structure and the underlying data. I realize that defining such criterion is by 
no means trivial. Thanks again for the help.

Best wishes,
Reza

Reza Khayat, PhD
Assistant Professor
City College of New York
Department of Chemistry
New York, NY 10031


From: Reza Khayat
Sent: Sunday, April 9, 2017 6:07 PM
To: CCP4 bulletin board
Subject: Structure comparison

Hi,

I have refined several structures of a protein from different space groups and 
would like to compare them to one another. Is there a program/software suite 
that would provide an objective comparison of the structures and identify 
regions where the structures are sufficiently different from one another to 
warrant a closer look? I think the most important aspect of the analysis would 
be defining a threshold (possibly based on resolution and structure statistics) 
that would identify sufficient difference between structures. Thanks.

Best wishes,
Reza

Reza Khayat, PhD
Assistant Professor
City College of New York
Department of Chemistry
New York, NY 10031

Re: [ccp4bb] DSSP output pdb format

[Gert Vriend]

Dear Adriana,

That server is ours, at the CMBI.

I realize that you are a bit confused about things. Feel free to address
me personally (and I am mailing this to the whole list because this
offer holds for all the 3D structure related servers reachable via
swift.cmbi.ru.nl/gv/facilities). In Skype I am gvriend and my mail
address is vri...@cmbi.ru.nl.

Greetings

Gert


On 4/3/2017 11:17 AM, Adriana Sene wrote:

Dear members
I assigned the secondary structure to my pdb file by the DSSP
application from this webserver
http://www.cmbi.ru.nl/dssp.html

The application works and gives output the pdb file.

The output file just have the reside list not the atom list. I am not
able to view the structure from this pdb file. If someone have
experience with this, or there is some other way to assign the
secondary structure to the pdb file.

best

Adriana


Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

Re: [ccp4bb] software or tool for Individual residue in a Ramachandran plot graph?

[Gert Vriend]

Ls,

It is not possible to determine the secondary structure from the 
position in the Ramachandran plot. For a residue to be helical, it must 
have hydrogen bonds that correspond with a helix, and so need its 
neighbours. If these conditions are met, a residue will by necessity 
have backbone torsion angles phi,psi that place it in the Ramachandran 
plot in the area known as helix.
However, if you look at a Ramachandran plot, you see that most residues 
are in the so called alpha helix, beta strand, or left-handed alpha 
helix regions. Actually, the percentage of residues that falls in these 
areas is the basis of the second oldest protein structure validation 
method. But on average almost half of all residues are in a loop. So 
where is the loop area in the Ramachandran plot? Well, there is no such 
area as residues in a loop tend to have in irregular succession of helix 
and strand torsion angles. So, if the torsion angles seem to suggest 
HHSHHHSSH then actually, we have a loop.
Due to the cooperativity of the hydrogenbonding patterns in regular 
secondary structure elements the torsion angles for a 
Ramachandran-helical residue are not the same when the residue is in a 
helix or in a loop, but those are details for the validation people, and 
will not displace the cross in the Ramachandran plot too much.


If you are interested in the secondary structure assignments, use DSSP. 
We made it open source and gave it to CCP4 already a while ago, so 
everybody can use it. DSSP has so its issues, but as everybody uses it, 
these issues are known, and softwares for which they are critical deal 
with those issues properly, normally.


Gert

On 03/24/2017 09:24 PM, Pavel Afonine wrote:

May be not exactly what you want, but should be close:

phenix.secondary_structure_restraints 1yjp.pdb

then here you have options

- for output format:

format = *phenix phenix_refine phenix_bonds pymol pdb refmac kinemage

- for search  (annotation method):

search_method = *ksdssp mmtbx_dssp from_ca cablam

I suggest you please with them and see which one suits you most.

Pavel



On Fri, Mar 24, 2017 at 1:18 PM, Xiao Lei > wrote:


Thanks Pavel,  is there a command that can tell secondary
structure assignment based on Rama plot of each residue beside phi
and psi? for example :
 A   2  ASN:56.93:-60.58:141.19:Favored:General alpha helix
 A   3  ASN:48.44:-119.25:125.15:Favored:General alpha helix

On Fri, Mar 24, 2017 at 1:09 PM, Pavel Afonine mailto:pafon...@gmail.com>> wrote:

Trivial using command line. Example:

- get a file from PDB:

phenix.fetch_pdb 1yjp

- get all phi/psi for all residues:

phenix.ramalyze 1yjp.pdb

residue:score%:phi:psi:evaluation:type
 A   2  ASN:56.93:-60.58:141.19:Favored:General
 A   3  ASN:48.44:-119.25:125.15:Favored:General
 A   4  GLN:16.23:-126.16:112.81:Favored:General
 A   5  GLN:55.13:-114.98:126.76:Favored:General
 A   6  ASN:6.17:-116.42:97.69:Favored:General
SUMMARY: 5 Favored, 0 Allowed, 0 Outlier out of 5 residues
(altloc A where applicable)
SUMMARY: 0.00% outliers (Goal: < 0.2%)
SUMMARY: 100.00% favored (Goal: > 98%)

Pavel


On Fri, Mar 24, 2017 at 10:37 AM, Nigel Moriarty
mailto:nwmoria...@lbl.gov>> wrote:

Alex

It seems that nobody has answered your question. I'm not
sure what you can do in CCP4, but if I understand your
question correctly, you can perform a comprehensive
validation in Phenix complete with Ramachandran plot
including clickable points relating to your residues which
allow you to see the residues in Coot.

Happen to help further on the PHENIXBB or off-line.

Cheers

Nigel

---
Nigel W. Moriarty
Building 33R0349, Molecular Biophysics and Integrated
Bioimaging
Lawrence Berkeley National Laboratory
Berkeley, CA 94720-8235
Phone : 510-486-5709 Email :
nwmoria...@lbl.gov
Fax   : 510-486-5909   Web 
: CCI.LBL.gov 


On Thu, Mar 23, 2017 at 8:39 PM, Alex Lee
mailto:alexlee198...@gmail.com>>
wrote:

Dear All,

Is there a tool or software which can give
Ramachandran information of individual residues in a plot?

I used Coot to check for Ramachandran plots, but it
shows all the residues in a coordinate I put in Coot,
not individual one. I also use "residue info" in coot,
it tells Ramachandran "phi psi" angles of individual
residue, but it does not show it in a plot, only numbers.

Thanks ahead for any input.

Re: [ccp4bb] 3D viewing room options

[Gert Vriend]

Hia Xavier,

For about 10 years already we use a simple solution consisting of a 
cheap computer with two graphics cards, a beamer from the game industry 
that can project stereo, and simple liquid crystal shutter glasses. This 
gives us projection on the wall (just a painted wall, nothing special 
with aluminum paint or so)  in a small room that we (with some sense of 
overstatement) call the cave. The stereo vision is excellent, and as the 
beamer and computer are next-door (beamer projects through a small 
double glass window), the room is silent. The image is nearly four 
square meter big, the cave-room is about 3 by four meters and sits 4 
people at one side of the table; they all have perfect stereo vision. In 
this whole set-up the most expensive part was getting the little window 
installed between the two rooms. It is my guess that if we had to 
install everything again today the costs would be clearly under 2000 
Euro (excluding the little window) perhaps even under 1000. The weak 
point in the whole setup are the glasses as they burn batteries like 
crazy. Our battery budget is probably more than 200 Euro/year.


If you want details, mail Elmar Krieger (el...@yasara.org); he installed 
it for us many years ago and he is THE specialist in this field


Greetings

Gert


On 12-11-2016 12:29, Brian Smith wrote:

Hi Xavier,

Any LG "Cinema 3D" capable TV should give you good passive stereo for the kind 
of thing you need and definitely work with PyMol and COOT under LINUX with a low end 
Nvidia Quadro (not NVS) card such as the K620 using the NVidia drivers.

Sadly LG seem to have dropped 3D support from their monitors, but many of their 
TVs have the capability with the smallest currently available in the UK being 
the 32 inch 32LF650V at ~GBP 400. I use a similar, older model on my desk as a 
monitor and it's just fine.

For a viewing room, a larger screen size version should be good and the 4K 
models should provide a big advantage. A better graphics card might be useful 
to drive them. I am trying to get a teaching classroom fitted out with e.g. the 
84UB980V at ~GBP 6,000. In terms of audience, the main issue with these types 
of screens is that you have to be at just the right height relative to the 
screen for the polarization filter on the screen to be aligned with the correct 
line of pixels, so make sure that you have adjustable height seats to cater for 
people of all statures!

I priced up the projector (beamer) option a little while back. For a small room, 
there's an EPSON product EB-W16SK that should do the trick easily for about GBP 
1,000. These are only WXGA resolution though so fine detail will be a struggle (do 
set line_width=3 in PyMol for example - this can be a good thing to do on the 
passive LCD screens too). Remember you'll need a projection screen that maintains 
the polarization (you can also get paint to do the job) - another £1,500 or so. For 
a bigger audience go to a specialist 3D vision company for a pair of projectors 
& polarizers with a budget of GBP 20,000+ in mind (mostly on the projectors).

We trialed an active 3D system also. The experience was not great - your 
position in the room seems to matter more than with a passive system - and many 
people reported motion sickness sensations - heavier glasses, flicker effect, 
etc. all probably exacerbate this tendency.

Brian Smith
---
IUPAB & EBSA Congress 2017 16-20 July 2017 http://www.iupab2017.org/
Institute of Molecular, Cell and Systems Biology & School of Life Sciences, 
University of Glasgow, Glasgow G12 8QQ, UK.
---

[ccp4bb] Trump

[Gert Vriend]

Lets stop this discussion before it divides this list of friends just as 
much as the cause of this discussion divided the US.


Gert

Re: [ccp4bb] Superpose program in CCP4

[Gert Vriend]

Dear Wenhe,

No 3D superpose tool will always align/map all Calphas. If in the one 
protein the loop turns left, and in the other it turns right, then 
mapping those loops is meaningless and thus not done by good software. 
The other problem is that often two proteins that get compared do not 
even have equally many residues so that there will always be some 
unaligned/unmapped Calphas left at the end. Look for some articles by 
Arthur M Lesk on this topic, he has explained protein superposition 
(problems) very clearly.


Gert

Ps, if you want proteins superposed and get different output from what 
the standard software gives you, just mail me those PDB files and I can 
see what I can do.



On 29-10-2016 17:47, WENHE ZHONG wrote:

Dear all,

I always use the SUPERPOSE tool in CCP4 to superpose molecules. This time I 
want to use the RMSD values of superposed C-alpha atoms to plot a RMSD graph 
(instead of using the graph automatically made by the program). However, there 
are many atoms missing in the RMSD list.

In the settings I chose “Superpose specific atoms/residues”, checked “Output 
all distances to a file”, fit “C-alpha atoms”. The superposed structures have 
exactly the same sequence.

My question is: is there any way to get the completed list of RMSD value for 
each C-alpha atom? Or is there any other program for this purpose?

Thank you!

Kind regards,
Wenhe

Re: [ccp4bb] Good 3D Monitor for Molecular Modelling

[Gert Vriend]

Hai Matt,

If you are not interested in hard-core crystallography stuff like 
interpreting electron-density maps or couplings to Coot or Refmac, then 
I can certainly recommend that you look at Yasara. We use it for 15 
years already for modelling and many different visualizations (and much 
more). I am especially happy with the fact that my entire WHAT IF 
software is included in Yasara. Yasara is one of the best modelling 
softwares available (Proteins. 2009;77 Suppl 9:114-22 
). Yasara is commercial, 
but very cheap, especially for academics. There is a limited version 
that is fully free of cost for everybody; called Yasara_view it does 
just enough to be very useful for education purposes. Yasara comes fully 
maintained and provides its customers a well-working help-desk.


(I do not benefit from Yasara sales...).

Look at www.yasara.org

Greetings

Gert


On 27-10-2016 17:20, Matthew Graf wrote:

Hello All,
 I am looking for suggestions on a good, but not too costly, 3D 
monitor for visualizing pdb structures and looking at outputs of 
modelling programs. I am not personally a structural biologist, but am 
on the hunt for someone who is. All help appreciated.


Kind regards,
Matt

[ccp4bb] Detwinning, why not?

[Gert Vriend]

Many, many years ago I worked in Rossmann's lab. Some colleagues were 
detwinning a structure. Whatever they tried, the twin ration went from 
something near 50-50 to something near 95-5 without any improvement. I 
wrote software that would look at reflection files in search of 
systematic absences (don't laugh these were the days before the 
internet; we still used puched cards) and corrected their spacegroup 
from P23 to P213 (or something similat; at least a sub-1 was added). 
That saved that day. So my only X-ray experience gave one example where 
detwinning doesn't work: wrong spacegroup.


Gert

Re: [ccp4bb] Hi

[Gert Vriend]

On 26/05/15 14:41, vijay srivastava wrote:

Dear All,

I want to superpose the nucleotide form one GTPase on to the 
nucleotide of other GTPase in order to study
the interaction in the nucleotide binding pocket. I tried to superpose 
but it is superposing on the basis of secondary structure as a
result both the nucleotides from two structutres are not properly 
aligned.  I want to superpose both  the nucleotide, so that I will get 
the matrix, which I want to apply on my desired strcuture and study 
the interacting residues.
Do any one have the align program with you or any other program which 
can solve this problem.


waiting for your kind response

regards
vijay


The WHAT IF software has an extensive superposition menu that allows you 
to superpose anything on anything as long as those two 'anythings' can 
be superposed (so don't superpose DNA on a protein helix...). 
Unfortunately you will have to go through the hassle of installing WHAT 
IF, and learning how to operate it also isn't trivial.


It is free, though.

Gert

Re: [ccp4bb] CYS mod

[Gert Vriend]

The site swift.cmbi.ru.nl/gv/numbers/ is not official, unpublished, and
still poorly maintained. But just for the fun of it, I added a list of
modified cysteines. Be aware that the files are big, so better
download them and read them in the editor than opening them in the browser.

Gert

On 05/12/2015 09:30 PM, David Schuller wrote:

What are the most likely modifications of a CYS residue? I am
attaching images of a couple of residues in my current structure. Blue
is 2Fo-Fc, Green is Fo-Fc positive difference, purple is model-phased
anomalous differences at 3 sigma, data from 1.70 A source.

I am guessing about 3 non-H atoms, probably one of them with anomalous
scattering.
Not all CYS in the structure are thus modified, only these two, but
they show up consistently in 4 noncrystallographic copies.

TIA,



Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

Re: [ccp4bb] [RANT] Reject Papers describing non-open source software

[Gert Vriend]

Being in software design myself, I want to look at it from the other 
side. If I produce software I want to get from it:

1) Gratitude from the users, preferably expressed in the form of citations;
2) Nice collaborations that allow me to use my, novel software at the 
edge of science;
3) The possibility to one way or another raise funds for future software 
design.
Point 1 normally goes OK, it is my estimate that, for example, WHAT IF 
is cited properly in 25-50% of all cases it is used, while for 
WHAT_CHECK this percentage is a bit lower. Point 2 sometimes goes OK, 
but point 3 really is a problem. I maintain WHAT IF and WHAT_CHECK using 
hospital money, which allows me to keep these facilities free of cost 
for you.


When I was the first application-note editor for the journal 
Bioinformatics, I defined a series of rules for software publications 
that included: 1) The software must work as advertised; 2) the software 
must be freely obtainable (to academics), or the webserver/webservice 
must be free to use; 3) There must be good Help facilities and examples 
that work as advertised; and 4, and this point is relevant for the 
present discussion, the software must remain available for at least five 
years (this has nowadays been shrunk to three years, I believe).


When I referee an application note these days, I look at the track 
record of the group that produced the software. If they published 
software over the past few years and that software is now gone, then I 
tell them to first get their old products back up and running and that 
they can resubmit their new software next year.


So, I agree with Robbie, but also with the many people who said that 
'we, ourselves' are the problem, not the system or the journals.


Greetings
Gert

Re: [ccp4bb] APBS

[Gert Vriend]

We addressed some of these electrostatics problems (Xray artefact like presence 
of ions, crystal packing, etc) a long time ago. Feel free to look at:

Improving macromolecular electrostatics calculations.
Nielsen JE, Andersen KV, Honig B, Hooft RW, Klebe G, Vriend G, Wade RC.
Protein Eng. 1999 Aug;12(8):657-62

and

Optimizing the hydrogen-bond network in Poisson-Boltzmann equation-based pK(a) 
calculations.
Nielsen JE, Vriend G.
Proteins. 2001 Jun 1;43(4):403-12.

Greetings
Gert


On 05/10/2015 12:37 PM, Andre Godoy wrote:
Dear users.
I'm comparing surface charge of a structure with its homologous (85% seq ID), 
and noted that APBS suggest completely opposite charge distribution (exactly 
what I expected, since despite its similarity molecules have opposite 
biochemical profile)

But since molecules are quite similar, I'm not convinced that charge 
distribution can be that different.
Considering that AA is basically the same, what other factors can be 
influencing APBS charge calculations? (Ex: rotamers position, crystal packing, 
crystallization conditions, etc)

best,


Andre Godoy
IFSC - University of Sao Paulo - Brazil


Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

Re: [ccp4bb] 3BDN, 16.5% Ramachandran Outliers!!!!!

[Gert Vriend]

At around 4.0 A resolution one normally cannot talk about accuracy. The 
density will at most locations not warrant any detailed interpretation. 
If, at 4.0 A resolution, you move atoms around a bit, you will not see 
significant changes in R/Rfree. So, you can do whatever you want, more 
or less. If the refinement puts great emphasis on the secondary 
structure (i.e. tries to force the Ramachandran plot, then you get a 
good Ramachandran plot, but if they put more weight of the fit to the 
density, you get a slightly lower R (and perhaps even Rfree) at the cost 
of the Ramachandran plot. And all of that is meaningless.
At this resolution you have to check if the fold is likely to be 
correct. I superposed just any high(er) resolution domain with high 
sequence similarity at the corresponding 3bdn domain, and I take any bet 
that the fold of 3bdn is right (at least of the domain I looked at, and 
I extrapolate to the other domain).
PDB_REDO (http://www.cmbi.ru.nl/pdb_redo/) cannot improve this one, and 
WHAT_CHECK (http://swift.cmbi.ru.nl/gv/pdbreport/) throws its hands up 
in the air. At 4.0 A you should be happy with an indication of the fold 
(and with the fact that the authors probably went through great pains 
for you getting these coordinates close to where they should be).


Gert

[ccp4bb] Renormalize a matrix

[Gert Vriend]

This matrix-renormalizer comes from WHAT IF. Feel free to use it any way 
you want:


  SUBROUTINE GVSREN (RMAT)
C+++
C---
C   

C RMAT IS A REAL MATRIX DIMENSIONED 
(3,3).  
C GVSEIG can be any eigenvalue calculator. Ask me if you want the 
WHAT IF one.
C   


C---
C===
  IMPLICIT  NONE
  INTEGER   I, J, K, L
  REAL  A(6), RMAT(3,3), S(3,3), T(3,3), X(3,3), Y
C
C FORM THE PRODUCT OF RMAT * RMAT(TRANSPOSE)
C
  L=0
  DO 30 I=1,3
 DO 20 J=1,I
L=L+1
A(L)=0.0
DO 10 K=1,3
   A(L)=A(L)+RMAT(I,K)*RMAT(J,K)
   10   CONTINUE
   20CONTINUE
   30 CONTINUE
C
C CALCULATE THE EIGENVALUES AND THE EIGENVECTORS OF A
C
  CALL GVSEIG (A,X,3,0)
C
C SMALL PRECAUTION AGAINST FUTURE OVERFLOWS
C
  A(1)=1.0/SQRT(AMAX1(A(1),1.E-12))
  A(2)=1.0/SQRT(AMAX1(A(3),1.E-12))
  A(3)=1.0/SQRT(AMAX1(A(6),1.E-12))
C
C FORM THE PRODUCT OF X * A
C
  DO 50 I=1,3
 DO 40 J=1,3
S(I,J)=X(I,J)*A(J)
   40CONTINUE
   50 CONTINUE
C
C TRANSPOSE X
C
  DO 70 I=2,3
 DO 60 J=1,I-1
Y=X(I,J)
X(I,J)=X(J,I)
X(J,I)=Y
   60CONTINUE
   70 CONTINUE
C
C CALCULATE THE RENORMALIZED RMAT AS S * X * RMAT
C
  CALL GVS3X3 (T,X,RMAT)
  CALL GVS3X3 (RMAT,S,T)

  RETURN
  END

[ccp4bb] strange pattern

[Gert Vriend]

The following article was rejected by 'our' journal...

http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)1097-0134/homepage/april_fools__day_special_papers.htm 



Greetings
Gert

Re: [ccp4bb] r.m.s.d. ΔB

[Gert Vriend]

Ls,

I dont think these numbers are useful by themselves. It makes more sense 
to look at differences in these numbers (between bound atoms in your 
protein and in 'the rest of the PDB') as function of as many external 
conditions as statistics allow. Then determine averages with SD, and 
score on Z and RMS Z.


Also, look at our recent BDB article in PEDS (Vol 27, Iss 11 457-462), 
because not every B-factor in the PDB is what you would expect it to be.


Gert

On 26/03/15 12:32, herman.schreu...@sanofi.com wrote:


Dear Bulletin Board,

A referee wants for the “Table 1” in the supplementary information the 
following data:


The r.m.s.d. ΔB (bonded atoms) (Å2)

All protein atoms

Main chain – Main chain

Side chain – Side chain

Main chain – Side chain

r.m.s.d. ΔB (Non-bonded contacts) (Å2)

All protein atoms

Using google I found at that some of these numbers could be calculated 
with Moleman, although I am not sure to what extend this program is 
still maintained.


Older versions of Refmac would calculate r.m.s.d. ΔB’s for main chain 
and side chain bonds, which I guess would be the “Main chain – Main 
chain” and “Side chain – Side chain” values requested. However, what 
would should I think of the “Main chain – Side chain” values; 
differences between Calpha and Cbeta atoms?


What would be the use of these numbers? The standard CCP4 validation 
programs, or any validation program I know, do not calculate these 
numbers, so they do not seem to be extremely important. If somebody 
could point me to a program which could calculate these number without 
too much effort, I would be happy to do it.  Otherwise, I would still 
be willing to go the extra mile if someone could convince me that it 
is useful to have these numbers.


Thank you for your help!

Herman

Re: [ccp4bb] looking for promotif...

[Gert Vriend]

On 03/06/2015 09:21 AM, Laurent Maveyraud wrote:

Dear all,

we are currently trying to pick up the trail of the PROMOTIF program
for which we can't find a link that works properly. Does anybody know
where I can find it ?
thanks for your help,

laurent

PROMOTIF seems to be an EBI product. Why don't you first try asking the
EBI? (It is a dead link at the EBI web site, but the EBI is an
international service institute so I would think they would respond
quickly if made aware of this problem.

Gert
Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

Re: [ccp4bb] checking the functionality of ccp4 for general research

[Gert Vriend]

Dear Li Xue,

The questions you are asking are more the domain of protein structure 
bioinformatics than X-ray crystallography. The CCP4 has a series of these 
programs available, but their collection is limited. May I suggest that you 
also look at http://swift.cmbi.ru.nl/gv/facilities/. There you can find 
pointers to a series of web servers that do things like you ask with PDB files 
as input. Most of the web servers are also available as a web service that you 
can call directly from your software. Further, several of the calculations you 
ask for we routinely perform once per week on all new PDB files and store the 
results in easy-to-parse files, one per PDB file.

Greetings
Gert
On 03/04/2015 11:20 AM, Li Xue wrote:
Hi,

I am new to ccp4. I am writing to check the possibility to use ccp4 as a 
platform to develop my software.

I was wondering whether I could get the following information from ccp4 after I 
input a pdb file:

1. chemico-physical properties of each residue, such as hydrophobicity, side 
chain charge, relative solvent accessibility and so on.
2. the nearest 5 neighbor residues of each residue in the pdb file
3. add hydrogen to the pdb file

Although I can calculate the above information myself without using ccp4, it 
would be great thatccp4 has already all the packages that I need. I'd 
appreciate it very much if someone could point out where I should start with.

Thanks.

Li


--
Xue, Li
Bijvoet Center for Biomolecular Research
Utrecht University
Email: l@uu.nl
Web site: http://www.cs.iastate.edu/~lixue/


Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

[ccp4bb] Ramachandran plot

[Gert Vriend]

I saw several excellent remarks about Ramachandran plots come by, but
the main points still seem missing, I think.

Phi and psi are not fixed parameters with limited freedom like angles,
bond lengths or planarities. By keeping a certain bond lengths
restrained at, for example, 1.543+/-0.021 Angstrom, you know what you
are doing, and you know that exceptions are very, very rare (assuming
that the weight on this restraint is set appropriately). Phi and psi,
though, can have very many values. And especially near beta-bulges or
near the ends of helices (or in active sites) one can easily do great
damage to the realism of the coordinates if phi-psi restraints would be
used in accordance with the local secondary structure.

The variability of a bond length or bond angle is determined by simple
physical, mainly local parameters such as through-bond and through-space
forces acting on the direct neighbour atoms, and this happens in a way
that we understand and that we can model in a coherent fashion. Phi and
psi angles, similarly, are greatly influenced by interaction of the four
atoms that determine each torsion angle with atoms in their direct local
environment. Robbie explained this nicely in his message. These forces,
though, are already restrained by the bond length, bond angle, and clash
avoidance restraints. If you now put in additional restraints for phi
and psi, you start counting certain interactions double in the
restraints. Unless you spend a lot of time on re-calibrating all
restraints, you are likely to generate areas where the sum of all
restraints is too high and areas where not enough restraints are left.
Again, I am not a crystallographer, but I am sure that that must make
R-free go up.

If using phi-psi restraints would make structures geometrically better
while at the same time make R and R-free go down, than we should surely
stimulate phi-psi restraints, but till this is proven, we shouldn't use
them. If one day the community concludes that phi-psi constraints are a
good thing, then please don't worry about the validation aspects. I am
sure we will come-up with novel validation methods.

Gert
Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.

Re: [ccp4bb] adjusting bad Ramachandran angles

[Gert Vriend]

On 25/02/15 18:08, Michael Murphy wrote:
Does anyone know of a way to adjust Ramachandran angles so that they 
fall within the preferred range? Either in Coot or possibly some 
online server? I have been trying to do it manually without much 
success, I was wondering whether there might another way to do it. -Thanks

Dear Michael,

Your question lacks the detail needed for a definitive answer, but I can 
imagine that you are solving a protein structure and have been told by a 
structure validation program that you have too many Ramachandran 
outliers. If so, then I would say, please don't fix it by adjusting phi/psi.


When you have an occasional headache, you take an aspirin or wait for 
the headache to go away, but when you have headaches every day, you go 
to a medic to search for the reason and get cured.


The same is with your structure. If you have one exceptional situation, 
you might decide to live with it, but when you have very many 
exceptional situations in your protein, you should not try to hide the 
problem by forcing some angles to fulfill our ideas about what things 
should on average be, but see how you can generally adjust/improve your 
refinement protocol to get a better match between the reality (measured 
reflections) and the model (coordinates).


I am not a real crystallographer, and you did not provide many details, 
so I cannot give any advice what you should do. But I can tell you what 
you should not do. You should not try to adjust phi and psi to pull 
residues into the Ramachandran plot's preferred regions.


Greetings
Gert

Ps, feel free to send me your coordinates (I promise to keep them 
secret) and I will see if the latest WHAT_CHECK provides some hints for 
what you can try to do.

Re: [ccp4bb] The discrepancy for determination of Ramachandran outliers by Coot and MolProbity

[Gert Vriend]

On 19-2-2015 1:38, Smith Lee wrote:

Dear All,
It often finds for the Ramachandran favored determined by Coot, MolProbity 
regards as Ramachandran outliers. There are earlier posts regards Coot and 
MolProbity has different database for the determination of the Ramachandran 
plots. Then will you please let me know the correct way to correct the 
Ramachandran outliers by Coot in order to meet the MolProbity Standards?
I am  looking forward to getting your reply.
Smith

Dear Lee Smith,

The ideas behind protein structure validation by usage of the
Ramachandran plot originally were to count outliers: J. Appl. Cryst.
(1993). 26, 283-291[ doi:10.1107/S0021889892009944 ] PROCHECK: a
program to check the stereochemical quality of protein structures R. A.
Laskowski, M. W. MacArthur, D. S. Moss and J. M. Thornton, or to
quantify these outliers: Objectively judging the quality of a protein
structure from a Ramachandran plot. Kleywegt and Jones (Structure Volume
4, Issue 12, 15 December 1996, Pages 1395–1400 Phi/Psi-chology:
Ramachandran revisited) have a few years later looked at phi,psi
combinations again, and tightened the Ramachandran plot contour-islands
a bit. The Ramachandran plot obviously has different contour lines for
different amino acid types. That is most obvious for proline and
glycine, but in the course http://swift.cmbi.ru.nl/teach/B2/ I explain
why the contour islands for Asp are systematically wider than for Glu,
for example. This idea (R.W.W. Hooft, C.Sander and G.Vriend, CABIOS
(1997), 13, 425-430) got implemented in WHAT_CHECK. MolProbity
(MolProbity: structure validation and all-atom contact analysis for
nucleic acids and their complexes Nucl. Acids Res. (2004) 32 (suppl 2):
W615-W619. doi: 10.1093/nar/gkh398) went back one step and used the old
ProCheck methodology, but they added the charm of colour bars.
These approaches, though, all are meant for validation purposes. Many
validation programs have been written, looking at many aspects of
protein structure quality, but the Ramachandran plot remained one of the
two most powerful validation methods mainly because it is hard to
optimize against the Ramachandran plot, and because the Ramachandran
plot is nicely reflecting very many aspects of protein structure
quality. If a protein structure producess a poor Ramachandran plot, then
you should not try to improve the Ramachandran plot, but you should try
to improve the structure. You can look if all ouliers fall in one loop,
and try to rebuild that loop, but in general, a poor Ramachandran plot
means that the whole structure has problems. Ramachandran plot quality
correlates nicely with resolution. So, if you have low resolution data,
you have little hope. You can use WHAT_CHECK to see how your structure
does relative to other structures that roughly have the same resolution.

If you want me to take a look at your structure, than please feel free
to send me the PDB file (and the mtz file), and we can take a look
(secrecy of coordinates guaranteed, of course).

Gert
Het Radboudumc staat geregistreerd bij de Kamer van Koophandel in het 
handelsregister onder nummer 41055629.
The Radboud university medical center is listed in the Commercial Register of 
the Chamber of Commerce under file number 41055629.