Messages

1351

General discussions (Q&As) / Standards on defining nucleic acid structural parameters

« on: January 10, 2012, 07:54:25 pm »

There are currently two consecutive standards on the nomenclature and definition of nucleic acid structural parameters, the initial 1989 Cambridge Convention, and the 1999/2001 Tsukuba Report (standard base-reference frame).

The 1989 Cambridge Convention
The 1989 EMBO Workshop (held at Cambridge, England) provided the first systemic "definitions and nomenclature of nucleic acid structure parameters" [Dickerson et al. (1989), Nucleic Acids Res., 17(5), 1797-803]. Thereafter, several analysis programs were updated or created (most notably Curves from Richard Lavery and NewHelix/FreeHelix from Richard Dickerson), following the loosely defined convention. Even though the parameters are now named the same, the different programs can provide conflicting numerical values, especially for strongly distorted structures. See, for example, Werner et al. (1996) "Intercalation, DNA kinking, and the control of transcription" [Science, 271(5250), 778-84].

The 1999 Tsukuba Workshop
Our efforts in "resolving the discrepancies among nucleic acid conformational analyses" uncovered the source of the disparate descriptions from commonly used software programs:

Growing interest in understanding the relationship between the global folding of nucleic acids and the sequence-dependent structure of individual base-pair steps has stimulated the development of new mathematical methods to define the geometry of the constituent base-pairs. Several approaches, designed to meet guidelines set by the nucleic acid community, permit rigorous comparative analyses of different three-dimensional structures, as well as allow for reconstruction of chain molecules at the base-pair level. The different computer programs, however, yield inconsistent descriptions of chain conformation. Here we report our own implementation of seven algorithms used to determine base-pair and dimer step parameters. Aside from reproducing the results of individual programs, we uncover the reasons why the different algorithms come to conflicting structural interpretations. The choice of mathematics has only a limited effect on the computed parameters, even in highly deformed duplexes. The results are much more sensitive to the choice of reference frame. The disparate schemes yield very similar conformational descriptions if the calculations are based on a common reference frame. The current positioning of reference frames at the inner and outer edges of complementary bases exaggerates the rise at distorted dimer steps, and points to the need for a carefully defined conformational standard.

This work laid the foundation for the standardization of the base-pair parameters in the 1999 Tsukuba Workshop on Nucleic Acid Structure and Interactions. Following public review and discussion, and final approval by the IUBMB Nomenclature Committee, the standard base-reference frame paper came out in 2001 [PDF from the Nucleic Acids Database (NDB)]. From the initiation to its completion, the standardization was made possible by the NDB project, led by Helen Berman.

While not a participant of the 1999 Tsukuba Workshop, I was nevertheless heavily involved in the preparation of the final report. In the following posts at this section, I am planning to provide full data files and scripts so the Figures and Tables can be re-generated. In addition to serving as learning materials for new comers to the field of nucleic acid structures, the details should also be of interest to seasoned practitioners to watch for the caveats. From my own perspective, the work is directly relevant to the automatic identification and classification of non-canonical base pairs in RNA structures, among other possible applications already implemented in or to be added to 3DNA.

To the best of my knowledge, currently only 3DNA and Curves+ conform to the standard base-reference frame; as a result, the structural parameters (for Watson-Crick base-pair duplexes) calculated by the two programs are very similar.

Related PDF Documents

• "Definitions and Nomenclature of Nucleic Acid Structure Parameters" -- report on the 'Cambridge Convention'
• "A Standard Reference Frame for the Description of Nucleic Acid Base-pair Geometry" -- report for the 1999 Tsukuba Workshop
• "Resolving the Discrepancies Among Nucleic Acid Conformational Analyses" -- report that laid the foundation for the standard base-reference frame
• "Overview of Nucleic Acid Analysis Programs" -- mathematical comparison of then seven commonly used software

1352

Users' contributions / Re: deformation energy calculation program

« on: January 07, 2012, 11:49:30 pm »

Please note that these two tests fail with the deformation energy program currently distributed on the 3DNA website at:

http://rutchem.rutgers.edu/~xiangjun/3DNA/users.html

The above note refers to the C++ program written by Marc Parisien (University of Montreal, Canada) for calculating deformation energies at the base-pair or dinucleotide step level. Since the referred to URL is now dysfunctional, I have consolidated all the related information here.

-- Xiang-Jun

Email message from Marc Parisien on the C++ program ('EnergyPDNA.C' and the associated header file 'EnergyParams.h' are attached below)

From parisien@iro.umontreal.ca
Date: Mon, 10 May 2004 11:28:18 -0400 (EDT)
From: Marc Parisien <parisien@iro.umontreal.ca>
Cc: xiangjun@rutchem.rutgers.edu, songliu@buffalo.edu
Subject: DNA deformation energy!


Hi All!

   I have received the mean values from Dr. Lankas;

        Lankas F, Sponer J, Langowski J, Cheatham TE 3rd.
          DNA deformability at the base pair level.
          J Am Chem Soc. 2004 Apr 7;126(13):4124-5.

   We now have the base-pair level deformation energies!!
   I have updated the program; see attached files :-)


   I have decoupled the energy calcs since there is the possibility
of coupled interactions between base-pairs and base-steps parameters...
(like Propeller-Rise, etc)   another article maybe for that ;-)
It would generate a huge matrix though: 12 x 12  (6 params for bases + 6
params for steps = 12 total params).

   I have also change the energy calcs loops from:

   1)
   i = 0 to 5  // the 6 parameters
   j = 0 to 5

   to

   2)
   i = 0 to 5
   j = i to 5

   because in 1) the i-j and j-i are counted twice, except for i-i,
   so it is not correct to then simply divide by 2.
   in 2) the i-j are counted singly, even for i-i.

   Regards,
   Marc.

//--------------------------------------------
Marc Parisien
parisien@iro.umontreal.ca
www-lbit.iro.umontreal.ca

Further note from Marc Parisien:

From parisien@iro.umontreal.ca Wed May 11 23:16:48 2005
Date: Tue, 19 Apr 2005 08:54:46 -0400 (EDT)
From: Marc Parisien <parisien@iro.umontreal.ca>
To: Bruno Contreras Moreira <contrera@ccg.unam.mx>
Cc: xiangjun@rutchem.rutgers.edu
Subject: Re: help with DNA deformation energy

Hi Dr. Bruno,


> I've just found your deformation energy code at
> http://rutchem.rutgers.edu/~xiangjun/3DNA
Cool!


> What's the energy reported? is it the energy that you need to apply to the
> system to obtain a given DNA deformation with a given sequence?
It is a deformation energy (units are kcal/mol, I think) based on
population preferences: E = -RT ln( P ) where P is the probability of
finding the sample in that conformation. This deformation energy reaches
back to Go where he analyzed the protein helix deformations (I don't
have that reference). I suggest that you imperatively read the 2 articles
mentionned in the "credits" section of the program (in the main()
routine).


> Do you need full-detail PDB coordinates of the DNA or the backbone is enough?
> Thanks for your feedback and for the code!!
Unfortunately, you need the full DNA since the energies come from the
side-chain conformations. You will have to launch the 3DNA program before
calculating the energies.


The energies reported here are not to be confused with those reported by
force-fields like AMBER or CHARMM... the energies here are those only from
population samples! You would have to do a thermodynamic cycle to obtain
the energy to apply to the system to obtain a given DNA deformation with a
given sequence!


As an application of this program you can look at a DNA/Protein complex
and do in-silico DNA mutations by changing the nature of the nucleotides
but without modifying the DNA 3D structure... You can then select the best
DNA sequence for that particular DNA 3D structure...


Do not mix the 2 energies calculations: use my program with the "-s"
option (the step energy) or with "-b" (the base-pairs energy)...



    Regards,
      Marc.

//--------------------------------------------
Marc Parisien
parisien@iro.umontreal.ca
www-lbit.iro.umontreal.ca

1353

Users' contributions / Building modified fiber models

« on: January 07, 2012, 11:26:32 pm »

This post is based on a series of email exchanges I had with Satoshi Yokojima (Institute of Materials Science, University of Tsukuba, Japan) in June 2004. This topic is on how to build base-only fiber models with varying rise, using a combination of various 3DNA components. Satoshi Yokojima kindly summarized the procedure from a user's perspective, as enclosed below.

-- Xiang-Jun

From yokojima@adenine.ims.tsukuba.ac.jp
Date: Fri, 25 Jun 2004 21:32:13 +0900 (JST)
From: Satoshi Yokojima <yokojima@adenine.ims.tsukuba.ac.jp>
To: Xiangjun LU <xiangjun@rutchem.rutgers.edu>
Cc: W. K. Olson <olson@rutchem.rutgers.edu>,
     Satoshi Yokojima <yokojima@adenine.ims.tsukuba.ac.jp>
Subject: Re: 3DNA --- building "modified" fiber models

Dear Dr. Lu:

I have finished the check of the new geometry and I found that
it is what I wanted to have.  The procedure you have suggested
worked nicely.  The thing what I did is as follows:

 [1] A new directory X3DNA/MY_Fiber55 is created.
 [2] The A|C|G|T.pdb files are copied from X3DNA/FIBER/Str55
    into X3DNA/MY_Fiber55.
 [3] In order to rebuilt the structure without backbones,
    I have deleted the backbone atoms from A|C|G|T.pdb
    using vi.
 [4] The following commands are executed in X3DNA/MY_Fiber55.
         > std_base A.pdb Atomic_A.pdb
         > std_base C.pdb Atomic_C.pdb
         > std_base G.pdb Atomic_G.pdb
         > std_base T.pdb Atomic_T.pdb
 [5] A new directory test/rise=10 is created where I wanted to
    produce the 55-th fiber model structure without the
    backbone and with Rise = 10 Angstrom.
 [6] The files Atomic_A|C|G|T.pdb are linked to the directory
    test/rise=10.
 [7] The 55-th fiber model structure 5'-GG-3' is generated in
    the directory test/rise=10 by the command:
         > echo "2"   > f55.in
         > echo "GG" >> f55.in
         > echo "1"  >> f55.in
         > fiber -55 f55.pdb < f55.in >& f55.out
 [8] The following command is executed.
         > find_pair f55.pdb stdout | analyze
 [9] The Rise value in the file "bp_step.par" is modified
    to 10 Angstrom by vi editor.
[10] The 55-th fiber model structure without the backbone
    and with Rise = 10 Angstrom is created by the command
    rebuild as follows:
         > rebuild -atomic bp_step.par f55_newRise.pdb
[11] The procedure from [5]-[10] is repeated for the different
    Rise values.

I think it is a good idea to add a lot of examples in the manual.
People can learn much easier from the casebook than from the
explanation of the commands.  I did not even imagine that the 3DNA
can do some of these things.  Also, if you have a good manual,
I can recommend other people to use 3DNA to build DNA structures.

It is very kind of you to explain the detailed usage of 3DNA.
Thank you very much, again.

 Sincerely yours,
 Satoshi Yokojima

    --------------------------------------
              Satoshi  Yokojima
      yokojima@adenine.ims.tsukuba.ac.jp
        Institute of Materials Science
            University of Tsukuba
    --------------------------------------

On Fri, 25 Jun 2004, Satoshi Yokojima wrote:

> Dear Dr. Lu:
> 
> Thank you very much for your quick reply.  It looks like it is
> working this time, but I think I need to do more careful check
> tomorrow.  I will send you another e-mail when I finish doing it.
> 
> Thank you very much, again.
> 
>  Sincerely yours,
>  Satoshi Yokojima
> 
>     --------------------------------------
>               Satoshi  Yokojima
>       yokojima@adenine.ims.tsukuba.ac.jp
>         Institute of Materials Science
>             University of Tsukuba
>     --------------------------------------
> 
> On Thu, 24 Jun 2004, Xiangjun LU wrote:
> 
> > Dear Satoshi:
> > 
> > Thanks for your message. I am so glad you noticed this slight difference.
> > Apparently, you are a rigorous scientist by not taking your tools blindly.
> > 
> > The reason for this slight difference is due to the fact that the standard
> > default base geometry (Atomic_A|C|G|T.pdb under X3DNA/BASEPARS) used by
> > 3DNA when rebuilding the structure is DIFFERENT from the geometry of the
> > fiber model 55 repeating units. 
> > 
> > Using a different set of base geometry for analysis and rebuild is
> > expected when we developed 3DNA, and the solution to your problem would be
> > as follows:
> > 
> > [1] create a new directory, say MY_Fiber55
> > [2] copy into MY_Fiber55 the A|C|G|T.pdb files from X3DNA/FIBER/Str55
> > [3] Since you would like the rebuilt structure WITHOUT backbones, you need
> >     to delete the backbone atoms from A.pdb etc manually, using any text
> >     editor (e.g., vi or emacs)
> > [4] run command 
> >               std_base A.pdb Atomic_A.pdb
> >       and repeat the above for C.pdb, G.pdb and T.pdb
> > [5] Then `analyze', change Rise in "bp_step.par", and `rebuild' as
> >     outlined previously, the "slight difference between the required and
> >     produced structure" will be gone.
> > 
> > Type "std_base" for more help information. Another issue to note here is
> > that `analyze' and `rebuild' check for standard base geometry files from
> > current working directory first, then those defined by environmental
> > variable X3DNA, and finally ~/X3DNA.
> > 
> > Please have a try and let me know what happens. It would help if you could
> > document step-by-step what you did. We are currently working to improve
> > documentations on 3DNA, and we certainly welcome users working examples.
> > Of course, we will properly acknowledge your contributions.
> > 
> > Best regards,
> > 
> > Xiang-Jun
> > 
> > On Thu, 24 Jun 2004, Satoshi Yokojima wrote:
> > 
> > > Dear Dr. Lu:
> > > 
> > > After examining the file produced by the method you have suggested,
> > > I have noticed that there is a slight difference between the required
> > > and produced structure.  The difference is in, for example, the bond
> > > length.
> > > 
> > >   I wanted to keep all the structural parameters in the 55-th
> > > fiber model but Rise.  It means I need to use the bond length and
> > > angles (including Hydrogen-bond) found in the 55-th fiber model.
> > > This is important because the 0.1 Angstrom difference of the bond
> > > length changes the single bond to the double bond.  Therefore, the
> > > results of the electronic structure calculations are quite different.
> > > 
> > >   Are there any method I can use to make 3DNA produce the structure
> > > I want to have?  Somewhat hard way, such as replacing BASEPAIR files,
> > > is fine for me, but I need to know what is the correct way to do it.
> > > 
> > >  Sincerely yours,
> > >  Satoshi Yokojima
> > > 
> > >     --------------------------------------
> > >               Satoshi  Yokojima
> > >       yokojima@adenine.ims.tsukuba.ac.jp
> > >         Institute of Materials Science
> > >             University of Tsukuba
> > >     --------------------------------------
> > > 
> > > 
> > > On Thu, 24 Jun 2004, Satoshi Yokojima wrote:
> > > 
> > > > Dear Dr. Lu:
> > > > 
> > > > Thank you very much for your detailed explanation about the coordinates
> > > > used in the fiber model.  Your suggested method to make the 55th fiber
> > > > model without a backbone and a different Rise value works perfectly.
> > > > That was just what I wanted to know.
> > > > 
> > > > Thank you very much, again.
> > > > 
> > > >  Sincerely yours,
> > > >  Satoshi Yokojima
> > > > 
> > > >     --------------------------------------
> > > >               Satoshi  Yokojima
> > > >       yokojima@adenine.ims.tsukuba.ac.jp
> > > >         Institute of Materials Science
> > > >             University of Tsukuba
> > > >     --------------------------------------
> > > > 
> > > > On Wed, 23 Jun 2004, Xiangjun LU wrote:
> > > > 
> > > > > Dear Satoshi:
> > > > > 
> > > > > Thanks for using 3DNA and your nice words about it.
> > > > > 
> > > > > Now, to answer your questions. Firstly, all the fiber models are based on
> > > > > literature work. Under directory X3DNA/FIBER, there is a README file, and
> > > > > a subdirectory for each of the 55 fiber models inlcuded with 3DNA. For the
> > > > > 55th fiber model, for example, you will find the following files at Str55/
> > > > > 
> > > > > ----------------------------------------------------------
> > > > >     A.pdb      C.pdb      G.pdb      T.pdb
> > > > >     A.rpt      C.rpt      G.rpt      T.rpt
> > > > >     TableIV.dat
> > > > > ----------------------------------------------------------
> > > > > 
> > > > > File "TableIV.dat" contains the data originally published by the authors.
> > > > > In this case, S. Premilat & G. Albiser "Conformations of A-DNA and B-DNA
> > > > > in agreement with fiber X-ray and infrared dichroism."  Nucleic Acids
> > > > > Research, 11(6), (1983), p.1897-1908.
> > > > > 
> > > > > A|C|G|T.rpt are the processed files in "pseudo"-PDB format, actually
> > > > > storing atomic cylindrical coordinates. These repeating units are used by
> > > > > 3DNA to generate the fiber models given a base sequence or the number of
> > > > > repeats. The corresponding *.pdb files are in real PDB format with the
> > > > > corresponding x-, y-, and z-coordinates converted from the cylindrical
> > > > > coordinates. These PDB files can be displayed using a molecular graphics
> > > > > program, such as Rasmol. While they are not directly used by 3DNA while
> > > > > generating the fiber models, they were helpful for "quality-control" in
> > > > > the initial process.
> > > > > 
> > > > > The 55 fiber model collected in 3DNA are from different authors, spanning
> > > > > several decades. Quite naturally, file formats, conventations for z-axis
> > > > > directions, atomic names etc, varied greatly, and of course, with typos,
> > > > > errors as well. To provide a useful tool to the community, we went through
> > > > > some quite painstaking procedures to make our collection as transparent
> > > > > and consistent as possible.
> > > > > 
> > > > > Secondly, with the above background information, and if I understand your
> > > > > question accurately, i.e. you would like a model based on fiber #55,
> > > > > without the backbone and a different Rise value. This could be easily done
> > > > > in 3DNA as follows:
> > > > > 
> > > > > [1] generate the 55-th fiber model using any base sequence you are
> > > > >     interested in, say named f55.pdb
> > > > > [2] find_pair f55.pdb stdout | analyze
> > > > > [3] modify file "bp_step.par" or "bp_helical.par" generated by [2] using
> > > > >     your favored text editor by changing the Rise values
> > > > > [4] rebuild -atomic bp_step.par f55_newRise.pdb
> > > > > 
> > > > > Hope this helps.
> > > > > 
> > > > > Xiang-Jun
> > > > > 
> > > > > On Thu, 24 Jun 2004, Satoshi Yokojima wrote:
> > > > > 
> > > > > > Dear Dr. Lu:
> > > > > > 
> > > > > > I am using your program 3DNA.  Thanks to your nice work,
> > > > > > it is very useful for my research.
> > > > > > 
> > > > > > Recently, I wanted to make the 55-th fiber model structure
> > > > > > but without backbone and give different Rise for an analysis
> > > > > > of the energetics of DNA.  Since I cannot do such a thing
> > > > > > by 3DNA as far as I understand, I tried to make a simple
> > > > > > program to do it.
> > > > > >  Then, I have noticed that I do not understand the frame well.
> > > > > > At first, I thought that the i-th local base-pair frame should
> > > > > > be used as a coordinate system to find Shift, Slide, Rise
> > > > > > between the i-th and (i+1)-th base-pairs translation.  However,
> > > > > > it looks like it is not the case.  Now I think that the helical
> > > > > > axis is taken as the z-direction of the coordinate system but
> > > > > > that's not enough for me to determine your coordinate system.
> > > > > > 
> > > > > > Could you kindly explain how you have taken the coordinate system
> > > > > > in 3DNA to calculate Shift, Slide, Rise.  If it is not so easy,
> > > > > > could you tell me any reference?
> > > > > > 
> > > > > > Thank you very much in advance.
> > > > > > 
> > > > > >  Sincerely yours,
> > > > > >  Satoshi Yokojima
> > > > > > 
> > > > > >     --------------------------------------
> > > > > >               Satoshi  Yokojima
> > > > > >       yokojima@adenine.ims.tsukuba.ac.jp
> > > > > >         Institute of Materials Science
> > > > > >             University of Tsukuba
> > > > > >     --------------------------------------
> > > > > > 
> > > > > 
> > > > 
> > > 
> > 
> 

1354

General discussions (Q&As) / Re: Problem of finding all base pairs in distorted DNA structure using 3DNA

« on: January 05, 2012, 06:21:52 pm »

Hi Difei,

Thanks for posting your question in the new 3DNA forum!

You've raised a very subtle, yet interesting and significant point w.r.t. to find_pair. There are several aspects to address the issue you experienced, as follows:

• There is actually one more (undocumented) 'sanity' check for base pair than those specified in the file 'misc_3dna.par': the overlap area between the two bases. The current fixed setting is 0.01 Ų, i.e., virtually allowing for no overlap. The 'missing pairs' in the two sample structures all have overlap areas over the criterion, which explains why you cannot find it by manipulating the various parameters in 'misc_3dna.par'. See attached figure for the pair A12_G with B13_C in 'md5.1004.mod.pdb' you emailed me.
• From my understanding of the term, I certainly won't classify such extreme cases as base pairs: base-overlapping is associated with stacking (vertical) instead of pairing (plannar). That's why I've not documented this criterion, but internally checked for it.
• To 'force' such pairs, it is best to first run find_pair on the structure (or a representative from an ensemble) and then make manual changes as necessary. The analysis routines in 3DNA, analyze/cehs, will calculate relevant parameters accordingly (indeed, of any arbitrary pair). The topic has shown up repeatedly in the forum, especially with regard to the analysis of NMR ensemble or MD trajectories. See the section "Molecular dynamics simulations".
• To allow for more flexibility,  it may be desirable (at least won't hurt) to have the overlap criterion also available in 'misc_3dna.par'. Do you want to try this? If so, please let me know your OS so I can provide you a test version.

HTH,

Xiang-Jun

1355

Feature requests / Re: Align multi-model (NMR) structures

« on: December 30, 2011, 01:34:31 pm »

The various ensemble related scripts will be consolidated in the coming v2.1 release.

Xiang-Jun

P.S. Did you know that one can click 'Notify' to receive email notification when a new post becomes available in a thread or forum you are interested in? It is a cool feature -- now I can be alerted with every new post or registration.

1356

Feature requests / Re: Align multi-model (NMR) structures

« on: December 28, 2011, 09:09:44 pm »

Hi Andrew,

Thanks for this feature request and for providing two sample images -- it would indeed make a nice feature for future release of 3DNA. There are many things I would like to consolidate/refine, and new features to add! This requested feature fits nicely to the overall scheme I have in mind.

Before I can get something for an "official" release of 3DNA v2.1 with this new feature included, I will try to come up with a Ruby script this week for you to try out.

Now to a specific point related to the procedure your outlined:

I have a suggestion on a new feature for multi-model NMR files, some sort of alignment command to align on each model's reference frame.  I performed this by hand as an example using the following

- Split the multi-model file into individual pdb files, 1 for each model
- for each model...
  - find_pair on the structure
  - rotate_mol using the ref_frame.dat file from find_pair

The pairing is consistent in all of the models but we have to run find_pair on each structure in order to get the reference frame data.

As with the MD simulation analysis script ('x3dna_md.rb'), it would be best for a user to specify the base pairing information. Thus the program 'find_pair' does not need to be run with each model. This would ensure consistency in all the models. Instead, the model-specific 'ref_frame.dat' can be generated with 'analyze'.

Check back ...

Xiang-Jun

1357

General discussions (Q&As) / Re: std_curved parameters seems not to work in 3DNA V2.1

« on: December 01, 2011, 11:25:51 am »

I certainly should have made such cases more explicit -- I will add a new section to the forum, titled "Bug reports and feature requests", by the end of the week. As I am working on improving 3DNA, do you have any desired features or suggestions to make the software more useful? Do not be shy!

Xiang-Jun

1358

General discussions (Q&As) / Re: std_curved parameters seems not to work in 3DNA V2.1

« on: December 01, 2011, 06:53:11 am »

Greetings Mauricio!

It was (still is) a bug, but see thread "Global Helical Axis Information Missing". It's well worths repeating that for a word-around (for the time being), please do the following:

Code: [Select]

<std_curved >0.6</std_curved >   # add a space in the end of the tag!Just to clarify, [red:298epu6t][mono:298epu6t]3DNA v2.0 [August 2007][/mono:298epu6t][/red:298epu6t] was a pre-release for testing of v2.0 (beta) before the 2008 3DNA Nature Protocols paper was published; [red:298epu6t][mono:298epu6t]3DNA v2.0 [June 8, 2008][/mono:298epu6t][/red:298epu6t] is the currently distributed version. 3DNA [red:298epu6t]v2.1[/red:298epu6t] is not out yet, but I am planning to release it in the near future -- by early next year!

Thanks for reporting the bug! As always, the more bug reports, the merrier!

Xiang-Jun

1359

General discussions (Q&As) / Re: PNA-DNA duplex study using 3DNA

« on: November 26, 2011, 10:06:26 am »

Hi SOUVIK,

Thanks for using 3DNA, and posting your question at the forum.

Your attached sample PDB file 'nspP.pdb' helped me trace the origin of the problem. As shown in the chemical structure of residue K19 in your PNA-DNA complex, its atomic naming clearly does not following the PDB convention for A (purine). See my blog post "PDB ATOM coordinates record". Thus the residue is not taken as a nucleotide by 3DNA (try: find_pair -s nspP.pdb stdout) to begin with, no wonder "no basepairs found".

To solve your problem, it is best to start from the source, i.e., the software (AMBER) which generates the PDB file. Alternatively, you can write a purpose-specific script to change the base names of each PNA residue following PDB naming convention for DNA/RNA. Thirdly, if there are enough interest and use cases, I may consider adapt or create such a converting script to be distributed with future releases of 3DNA.

HTH,

Xiang-Jun

1360

General discussions (Q&As) / Re: modelling ssDNA or ssRNA

« on: November 17, 2011, 07:38:23 am »

Hi Laura,

Thanks for your interest in 3DNA and for posting on the forum.

I understand what you mean, but I do not think 3DNA (or other tools I know of) can provide you with a straight answer. It all depends -- what specifically is your definition of "an extended conformation", i.e., "single strands without helical configurations"? If you have a starting model, mutate_bases and analyze/rebuild/frame_mol in 3DNA may help.

As always, I am willing to extend 3DNA in ways that make sense to me and are useful to the community. Currently, 3-dimensional nucleic acid structure modeling is one aspect I am interested in.

Xiang-Jun

1361

General discussions (Q&As) / Re: DNA Triple Helix Generation

« on: October 22, 2011, 09:46:05 am »

Hi Himanshu,

I know English is not my native language, but do you really have any problem in understanding my previous two replies to your questions in the thread?

I wish 3DNA could be helpful to your project. However, 3DNA has been designed for some specific purposes, and apparently it does not fit your need. Please check some other (docking) software, e.g., HADDOCK from Alexandre Bonvin. Good luck!

Xiang-Jun

1362

General discussions (Q&As) / Re: DNA Triple Helix Generation

« on: October 21, 2011, 07:18:38 am »

Hi Himanshu,

Fig1.png(as attachment) : ... green ribbon shows single strand DNA which is ligand to dock in double stranded DNA. In the green ribbon there is intra-molecular bonding.

Thanks for attaching a PNG image illustrating the unusual "intramolecular bonding(backbone and nitrogen bases)". It is certainly clearer than pure text description. However, it would be even better if you provide the corresponding PDB file so others can easily visualize the problematic structure.

Can we ignore those bonds?

I do not know how you generated the structure, and for what purpose, thus I cannot offer any advice in this regard. The question appears not to be directly 3DNA-related. However, the image does show that the ssDNA is in poor geometry.

Xiang-Jun

1363

General discussions (Q&As) / Re: DNA Triple Helix Generation

« on: October 20, 2011, 07:01:55 am »

Hi Himanshu,

Thanks for posting at the 3DNA forum. Just to be clear, 3DNA does not perform DNA-DNA docking or any energy calculations.

But, the ssDNA show intramolecular bonding(backbone and nitrogen bases) which is un-usual.

Could you elaborate on it? Better with a specific example.

However, 3DNA does provide the following approach for DNA triplex creation which may be useful to your project:

• Build a fiber model (e.g., #31) for Poly (U) : poly (A) : poly(U) (11-fold). Type "[mono:26cqmn7j]fiber -m[/mono:26cqmn7j]" for more info, and search the forum for related posts. Alternatively (or better), pick a related triplex structure from the PDB, or another source.
  
• Perform base mutations (using [mono:26cqmn7j][red:26cqmn7j]mutate_bases[/red:26cqmn7j][/mono:26cqmn7j]) to your specific sequence. See thread "change one base pair in a double-strand DNA structure file".

Please note that the mutated triplexes may have steric clashes (atoms too close) or longer than ideal H-bonding geometry: they are intended as initial structures for refinement using any of the available energy calculation tools.

HTH,

Xiang-Jun

1364

General discussions (Q&As) / Re: 3DNA on 64 bit Linux (Ubuntu)

« on: October 19, 2011, 12:59:01 pm »

I have a problem with running 3DNA on 64 bit Fedora. It says "can't execute the binary file" whenever I try to run any of the programs from 3DNA/bin directory. The PATH and X3DNA environmental variables are set correctly.
Is there a version of 3DNA for 64 bit Linux systems available?

Glad to know the detailed error message. With two cases, now it seems clear that the 32 bit version of 3DNA compiled in Ubuntu does not work in 64 bit machines.

Check email for download info of a 64-bit compiled version of 3DNA and report back if it work on your 64 bit Fedora.

Thanks for using 3DNA and posting in the forum.

Xiang-Jun

1365

MD simulations / Re: H-bonding information in MD analysis output

« on: October 04, 2011, 07:25:42 am »

Hi Andre,

I will try to work with the extract output and organise the data as you suggested. I guess I'm trying to make it easier to see the type of H-bond events that are occuring.

It makes sense. Over the years, 3DNA has been used in ways that are out of my original intention. As a principle, I won't add a 'feature' into 3DNA unless I fully understand it and can back it up when asked. Put another ways, feature-rich is not what 3DNA aims for, but quality is. This is where purpose-specific scripting comes into play, and such a skill is essential for a 'bioinformatican'.

I am not experienced in programing/scripting so it will be an interesting challenge. I'll post the script here if I'm successful as others may have a use for it.

We all learn from practice, and by making mistakes. I am willing to help you in the exercise: just put your script with a reproducible example here, then I (hopefully other viewers of the forum) can make suggestions. With several iterations, the problem will become clearer, and your script may well function for your needs. In retrospect, this is the most effective way to learn things, and I wish to have been guided this way. I aim to turn this forum into a 'educational' resource so that people with similar mind can share and benefit from each other's experience and contributions. I am glad that you are willing to share, which is precisely the incentive that I'd like to help you more.

Have a try, and let's see how it goes.

Xiang-Jun

1366

MD simulations / Re: H-bonding information in MD analysis output

« on: October 03, 2011, 10:19:00 pm »

Hi Andre,

This may be a process that the extract script will have to handle after 'analyze'.

As mentioned previously, I am not convinced to add such a functionality directly in the current two Ruby scripts. If you want to proceed, it is preferable (in my understanding) to start from the extracted [mono:13mhwi7u]<hbond>[/mono:13mhwi7u] parameters. A purpose-specific script is the way to go.

Let's use wastson crick pairing as our conventional H-bond information. For your GC pair example, can we extract and organsie the information so the first column is O2 - N2, second is N3 - N1, and the third is N4 - O6. If there are non-waston crick H-bonds they can be placed in the fourth and subsequent columns.

In addition to G-C pair, you also need to consider C-G pair, which by default swaps the order of atoms in H-bond output.

Do we have to specify all the non-waston crick H-bonds or can the script look for the waston crick H-bonds and place the non-waston crick H-bonds after the standard waston crick information?

It is entirely up to you, with a purpose-specific script.

HTH,

Xiang-Jun

1367

MD simulations / Re: H-bonding information in MD analysis output

« on: October 02, 2011, 09:40:40 am »

Since the H-bond information for each base pair is arranged in columns, is there a way for the script to define each column for a specific H-bond and if the bond does not exist then it will leave it blank?

The parsing script has no knowledge of any 'specific H-bond': it just takes H-bonds for each base-pair as a single unit of text string. The H-bond info is calculated with 'analyze', based on purely geometric criteria. Thus for example there can be different A-T pairs: e.g., Watson-Crick pair and Hoogsteen pair, and many other possibilities.

In the example shown in my first reply, note the first two C-G pairs:

1 C-----G  [2]  O2 - N1  2.99  N4 - O6  2.98
2 C-----G  [3]  O2 - N2  2.74  N3 - N1  2.84  N4 - O6  3.17

They are based on model #20 of the distributed PDB ensemble 'sample_md0.pdb'. Note specifically, while the 2nd C-G pair is normal with 3 H-bonds, the first C-G pair has 2 H-bonds including the unconventional 'O2 - N1  2.99'.

Overall, I cannot see a sensible, generally applicable way to "define each column for a specific H-bond".

Here is a file to show you what I mean

I realise this is just formatting however it will make life easier to have the script format the output rather than do it manually.

See above. For your example, the first column does not align. Think about all the possible base sequence and pairing patterns, what you are asking for is certainly not just a formatting matter.

0   [3]   O2 - N2   2.96   N3 - N1   3.03   N4 - O6   3.49
1   [3]   O2 - N2   3.53   N3 - N1   2.97   N4 - O6   2.75
2   [3]   O2 - N2   3.01   N3 - N1   2.78   N4 - O6   3.19
3   [3]   O2 - N2   3.17   N3 - N1   3.25   N4 - O6   3.53
4   [3]   O2 - N2   2.81   N3 - N1   2.97   N4 - O6   3.22
5   [3]   O2 - N2   2.86   N3 - N1   2.87   N4 - O6   2.86
6   [1]         N3 - N1   3.08      
7   [2]         N3 - N1   3.25   N4 - O6   2.78
8   [2]   O2 - N1   3.07         N4 - O6   3.35
9   [3]   O2 - N2   2.96   N3 - N1   2.71   N4 - O6   2.85
10   [3]   O2 - N2   3.32   N3 - N1   2.87   N4 - O6   3.09
11   [3]   O2 - N2   3   N3 - N1   2.86   N4 - O6   3.18
12   [2]         N3 - N1   3.14   N4 - O6   3.04
13   [2]   O2 - N2   2.75   N3 - N1   3.29      
14   [3]   O2 - N2   3.04   N3 - N1   3.06   N4 - O6   3.38
15   [3]   O2 - N2   2.85   N3 - N1   2.99   N4 - O6   3.21
16   [3]   O2 - N2   3.17   N3 - N1   2.93   N4 - O6   2.98      
17   [3]   O2 - N2   2.79   N3 - N1   2.74   N4 - O6   2.6      
18   [3]   O2 - N2   3.03   N3 - N1   2.78   N4 - O6   2.66      
19   [2]   O2 - N1   2.75               N3 * O6   3.32
20   [2]   O2 - N1   2.99         N4 - O6   2.98

I understand you may have a clearly defined goal in hand, and that's where application-specific scripting/programming comes into play. I may be able to help in some way if you make your case more specific.

Xiang-Jun

1368

MD simulations / Re: H-bonding information in MD analysis output

« on: September 30, 2011, 10:17:44 am »

Hi Andre,

Please check the updated post "Ruby scripts for the analysis of MD simulation trajectories". Now in v0.7, the scripts parse and extract H-bonding information you asked for, and also the overlap areas for quantifying stacking interactions.

Now I realized that I had a (solid) reason to exclude H-bond parameters in the scripts -- being variable-length text instead of float numbers (see my first reply in this thread for an example), they do not fit the pattern with other previously extracted parameters. Specifically, with N models (structures) and each M base-pairs, we have a N-by-M data matrix for each of the base-pair parameters.  For buckle, see the attached file 'buckle.tsv' (based on the distributed dataset [mono:kldkpnjg]sample_md0.pdb[/mono:kldkpnjg]. Here '[mono:kldkpnjg]tsv[/mono:kldkpnjg]' stands for tab-separated values). Now for H-bonding, the extracted data file 'hbond.tsv' looks a bit wildly -- each item has different text length and number of H-bonds. I cannot see how such information can be processed directly as with other parameters, and I'd be interested in knowing how you proceed.

Have a try, and see how it goes.

Xiang-Jun

1369

MD simulations / Re: H-bonding information in MD analysis output

« on: September 29, 2011, 07:03:12 am »

Hi Andre,

Thanks for your feedback. I will add a section of H-bond parameters (hopefully) by tomorrow, and then release an updated version of the Ruby scripts. So stay tuned, and check back the forum.

Xiang-Jun

1370

MD simulations / Re: H-bonding information in MD analysis output

« on: September 28, 2011, 07:17:37 am »

Hi Andre,

First, thanks for posting your question in the forum! The more user feedback, the merrier.

Second, you may notice that I have split your post from the thread "Ruby scripts for the analysis of MD simulation trajectories" which is too long (18 posts now), and spans two pages.  I have added a new subject line "Re: H-bonding information in MD analysis output" for the new thread.

Third, and more relevant to your question, are you referring to adding a new section for the H-bonding information? An example would be:

<hbond>
   1 C-----G  [2]  O2 - N1  2.99  N4 - O6  2.98
   2 C-----G  [3]  O2 - N2  2.74  N3 - N1  2.84  N4 - O6  3.17
   3 T-----A  [2]  N3 - N1  3.09  O4 - N6  2.86
   4 A-----T  [2]  N1 - N3  2.78  N6 - O4  3.05
   5 A-----T  [2]  N1 - N3  2.81  N6 - O4  2.76
   6 T-----A  [2]  N3 - N1  2.71  O4 - N6  3.06
   7 A-----T  [2]  N1 - N3  3.12  N6 - O4  3.12
   8 G-----C  [2]  N1 - N3  3.00  O6 - N4  2.75
   9 A-----T  [2]  N1 - N3  2.93  N6 - O4  3.12
  10 A-----T  [2]  N1 - N3  2.90  N6 - O4  2.93
  11 A-----T  [2]  N1 - N3  3.01  N6 - O4  3.05
  12 T-----A  [2]  N3 - N1  2.80  O4 - N6  3.12
</hbond>

If that's the case, then it won't be much a problem to parse and add it into the output file. Otherwise, please provide a concrete example to show exactly what you mean.

Please confirm.

Xiang-Jun

1371

General discussions (Q&As) / Re: constructing a modified sugar

« on: September 12, 2011, 08:21:52 am »

...i am successful in making the modification using chimera software which has a menu called 'Build structure'. It would be nice to have this feature incorporated in 3DNA.

Glad to hear that you are successful in making sugar modification with Chimera 'Build structure'. If that does a decent job, why should 3DNA bother to compete (recreating the wheel)? That's exactly why I asked you the question in my preview reply.

Xiang-Jun

1372

General discussions (Q&As) / Re: 3DNA on 64 bit Linux (Ubuntu)

« on: September 09, 2011, 11:19:58 pm »

Hi Fei,

I did not realize 3DNA v2.0 compiled in 32-bit Ubuntu Linux does work in a 64-bit machine. Yours is the first reported case. Any comment / experience from other users?

Anyway, check email for download link to 3DNA v2.0 compiled in Scientific Linux 6 x86_64. Do report back here if it works, or not.

Xiang-Jun

1373

General discussions (Q&As) / Re: constructing a modified sugar

« on: September 08, 2011, 05:14:02 pm »

Hi Subbu,

Thanks for posting this question. While I understand what you mean, I must say that 3DNA (in its current version) does not have direct means to get the job done. I may be able to come up with a solution after reading the publication associated with PDB id 2X2Q, check the PDB coordinates of the entry, and then play around with 3DNA to fit the bill. Obviously, this will take time.

To make this effort worthwhile, I'd ask you to perform a survey of other tools, e.g., UCSF Chimera, PyMol etc, to see if they already provide similar functionality. Of course, others viewers of this thread are more than welcome to jump in with their opinions. Hopefully, more users will get actively participated in such discussions that will help shape further development of 3DNA.

Xiang-Jun

1374

General discussions (Q&As) / Re: building a parallel stranded DNA

« on: September 07, 2011, 01:30:51 pm »

Hi,

Thanks for attaching the three PDB files -- they indeed clarify the issues.

Do you know the rationale of the no preserve option? It appears there can be many ways to rotate around the glycosidic bond to orient the base.

Xiang-Jun

1375

General discussions (Q&As) / Re: building a parallel stranded DNA

« on: September 07, 2011, 12:42:16 pm »

Hi Bala,

1) Firstly i tried by constructing the ps DNA from fiber model 31 but the structure i obtained was nt correct (in the sense the base pairs of chian A and C) were not in proper orientation. So i used the pdb structure as template to carry out mutation.

It seems that [red:27t8jes7]mutate_bases[/red:27t8jes7] serves your needs, right?

For the record, in fiber model #31, chains A and C form a parallel duplex with classic Hoogsteen A-U pairs. 3DNA has no problem in analyzing such structures. See the attached structure for an example:
[attachment=0:27t8jes7]fiber31_Hoogsteen.pdb[/attachment:27t8jes7]
[pre:27t8jes7]****************************************************************************
Structure classification:

This is a parallel duplex structure
****************************************************************************[/pre:27t8jes7]

2) I am not getting what example you need, in case if it is related to chimera software, you can refer this page. It is freely down loadable.

When you talk about the '[red:27t8jes7]swapna[/red:27t8jes7]' command with '[red:27t8jes7]preserve[/red:27t8jes7]' set to TRUE or FALSE, it would be helpful to be concrete by using examples. More specifically, you start with a structure in PDB format, and try to mutate a base to another one -- you have two results by setting '[red:27t8jes7]preserve[/red:27t8jes7]' to either TRUE or FALSE. If you attach these three PDB files, then things would be much clearer even for those who are not familiar with the UCSF chimera software.

Xiang-Jun