Messages

1526

w3DNA -- web interface to 3DNA / Re: Jmol as an alternative for visualization?

« on: December 09, 2008, 04:57:14 pm »

Regarding to the fiber model construction, Wilma suggested generate a fiber with combination of different fiber models, such as 5 repeats A form + 5 repeats B form. We understand that the difficulty is the connection of two fiber models (backbones). What do you think of having this option on our web server?

I understand Wilma's point, but I honestly do not think it is a good idea here. There are could possibly be many different combinations of the various fiber models, which could be very creative, but  arbitrary. This is exactly what I want to avoid in this work, but it could possibly be left to another paper.

Basically, this is not a scientific paper, but a web-interface to commonly used functionality in 3DNA, and we've already had many stuffs to offer. Additionally, as a general design principle, the web-interface should be simple, targeting the non-expert users. Power-users will play directly with the command-line version.

Xiang-Jun

1527

w3DNA -- web interface to 3DNA / Jmol as an alternative for visualization?

« on: December 09, 2008, 02:35:29 pm »

Hi Guohui,

I noticed that you are using WebMol for visualization, which is fine. However, you might consider to add Jmol as an alternative, at least. Jmol seems to be quite popular right now with an active community. I have recently asked a question regarding Jmol support on alchemy file format there, and received near ten responses.

Also, for the fiber models, I thought we could have a page with all 55 models each with images in (three) different views, citations to the original references etc. This will give users a direct impression on the comprehensiveness of possibilities available with w3DNA/3DNA. We certainly have no matches, as far as I am aware of. A handy (robust and efficient) fiber-model generation and visualization service will by itself be a useful tool to community at large (e.g., for educational purpose), as I mentioned to you before.

Thanks for answer questions in the open 3DNA forum. You will realize that it is good thing to do: just keep doing it!

Xiang-Jun

1528

w3DNA -- web interface to 3DNA / Re: Logo

« on: December 09, 2008, 12:30:12 am »

Hi Guohui,

Good progress!

Try to finish 'fiber' model building first, as suggested earlier. Present the user with a full-list of available models (55), and based on selected model number, ask for sequence or just repeating unit. It is fine to have the most common A-, B-, C- and Z- (not D-, or with D-) form at the top.

I am not a designer either -- in the 3DNA forum, I have been asking for user's contributions. Until very recently have I updated the 3DNA forum logo. But honestly, I do not like the one you currently post there. I have created an alternative as attached here, which has a transparent background.

For 'analyze', you could pre-build all available NDB entries. Thus when a NDB/PDB id is supplied, the results are immediately available.

Keep good work!

Xiang-Jun
[attachment=0:pn02lib2]w3DNA_logo.gif[/attachment:pn02lib2]

1529

General discussions (Q&As) / Re: Continuous polymer

« on: December 03, 2008, 11:05:08 pm »

This is a good question! Unfortunately, 3DNA do not have a direct answer for it (yet). Does anyone know of a general solution to this problem (say, from a commercial or freely available software package)?

In principle, though, one would image to keep each repeating unit (8-bps here) as a rigid body and perform transformations (rotations + translations) to get the next and so on. One way to achieve it is by properly setting coordinate frames.

Xiang-Jun

1530

w3DNA -- web interface to 3DNA / Re: Rebuilding part

« on: December 01, 2008, 01:45:18 pm »

For the fiber model-building part, I am trying to use the program "fiber", am I right? This program requires parameter inputs from the screen/keyboard, to provide the sequence information. However, if we want to execute it from the web, this could be a trouble, since we won't be able to type it those information. Is it possible that you can adjust the program a little bit such that all parameters (maybe data files) could be included in just a command line, such as "fiber -s sequence.txt -a fiber_a.pdb"?

Right. User needs to first select a model, which can be provided via a HTML form selection list. Depending on the model selected, the sequence could be fixed, where user just needs to select the number of repeating unit (form input field with type="text"), or user need to input a sequence. The sequence could be from a file which is to be uploaded to the server, or through textarea tag. All such info is collected from a web-based form, and then passed to fiber. The program needs not to be changed, by taking advantage of Unix file indirection, e.g., '<' or '>'. Please play around in command-line first to get it right before you are able to implemented a web-interface to it.

Or do you have any other ways to solve this? Can I use "rebuild" function as an alternative? How?

The rebuild program is for a different purpose, thus it is not a substitute for fiber. Let's get fiber functionality done first: itself would be of great value to a large audience.

HTH,

Xiang-Jun

1531

w3DNA -- web interface to 3DNA / Re: What kind of services we will provide?

« on: November 30, 2008, 01:32:06 pm »

Hi Guohui,

Good start!

Overall, I would like to make w3DNA simple, accurate and robust, at least at the very beginning: take, for example, the 'google' home page. Do not put too many stuffs; whatever is there should be useful.


• Analyzing: list the various 3DNA output files files for download first, followed by any extracted info. A nice, simple add-on would be to mark non-Watson-Crick pairs (or G.U wobble pair).
• Rebuilding: put fiber model-building part first since this is the most-useful part to the general community; add sequence-specific building funcationality next: do not forget with sugar-phosphate backbone in various conformation, and the simplified Calladine-Drew style in Alchemy format. The download page should also be linked to Jmol/RasMol for online view.
• Visualizing: blocview is certainly the first choice, followed by stacking diagram, and base-pairing diagram (see the 3DNA NP paper). This is linked directly to the analyzing part.

You might want to put fiber-model building first in the list, following by visualization. In the home-page, you might also want to put some nice images to illustrate 3DNA major functionality to attract visitors.

Best regards,

Xiang-Jun

1532

General discussions (Q&As) / Re: ideal values for the stacking parameters

« on: October 10, 2008, 11:03:10 pm »

Having never performed MD simulations, I am not sure how an idealized uniform structure would effect your comparisons. With 3DNA, you can build a structure with any prescribed step parameters. However, the backbone geometry would normally be distorted, especially at the connections between nucleotides.

What initial structure do you start with for your MD? Can't it be used for your comparison? How about fiber B-form DNA? Any comments from MD experts?

Xiang-Jun

1533

General discussions (Q&As) / Re: Calculating the angle of DNA curvature

« on: September 26, 2008, 11:47:40 pm »

Thanks for reading the 3DNA Nature Protocols 2008 paper (NP2008).

The quota you cited refers to the section titled "Relationship to other programs", more specifically the "-c" option of "find_pair" designed to make Curves users' life a bit more straightforward. Curves is certainly a well-known program in quantifying DNA curvature, as evidenced clearly in literature where DNA bending angles are reported.

In the 3DNA NP2008 paper, protocol (recipe) no. 4 is on "Automatic identification of double-helical regions in a DNA–RNA junction", which provides detailed steps on calculating the angle between helices #1 and #3.  If that's what you want, you may find it worthwhile to (re)read the relevant parts more carefully.

Given the many similar questions recently popped up in the forum, it seems fair to say that quantifying DNA bending angle or curvature is still an open issue,  at least not well-understood by non-experts in the community. From my experience (not just in nucleic acid structures), this is not surprising at all. A seemingly long-solved problem still could bug you down when you try to get to the bottom of it.

HTH,

Xiang-Jun

1534

General discussions (Q&As) / Re: local helical parameters interpretation question

« on: September 18, 2008, 12:02:51 am »

That's correct.

As a side note, the helical axes in Figure 4 of 3DNA 2003 NAR paper is actually 11 - 1 = 10 fragments which are perfectly aligned in the regular structures.

HTH,

Xiang-Jun

1535

General discussions (Q&As) / Re: interhelical angles

« on: September 16, 2008, 09:13:47 pm »

Unfortunately, the simple answer is NO.

However, you might want to refer to the 3DNA Nature Protocols paper recipe no. 4 on "Automatic identification of double-helical regions in a DNA–RNA junction" and explanations therein on how this problem is handled in not "a straightforward way" with 3DNA. Browsing the forums should also give your some hints.

HTH,

Xiang-Jun

1536

General discussions (Q&As) / Re: local helical parameters interpretation question

« on: September 16, 2008, 09:05:15 pm »

It is not a surprise to me that the local helical parameters as given in 3DNA could be a bit confusing. I have answered this question several times over the years, mostly before the forum was set up. Excerpted below is one I just dug out:

[1] To refer the orientation and position of one base-pair (bp) relative to
    the other, 6 parameters (3 rotations and 3 translations) are required.
    One set of such parameters is (Shift, Slide, Rise, Tilt, Roll and
    Twist), and the other set is (X-displacement, Y-displacement, Helical
    Rise, Inclination, Tip and Helical Twist).

    Obviously these two sets should be completely reversible/dependent: from
    any one set you can get the other, rigorously. You can verify this point
    using "step_hel", a utility program in 3DNA. Graphically this is best
    illustrated by the Calladine-Drew A to B transition model by introducing
    uniform Roll and Slide values at each dinucleotide step. You could see
    these images in 3DNA website, Examples/Calladine_Drew/ directory in 3DNA
    distribution and 3DNA user's manual. The key point is that by
    introducing Roll, you also get Inclination, and with Slide, you get
    X-displacement.

    The "rebuild" program in 3DNA can construct a DNA structure using either
    set of these parameters. Examples of such input files (e.g.,
    "bp_step.par" and "bp_helical.par") can be generated by
    "analyze". Please have a look of the Examples/Analyze_Rebuild directory.

[2] The define a local helical axis, we need two base-pair reference
    frames (i and i + 1). 3DNA finds the single-helical axis (which is
    actually dx times dy) that brings i to coincide with i + 1 by a Helical
    Twist angle. The position which this helix passes through is defined by
    Chasles' theorem as detailed in Figures 12 & 13 of Backcok et al.
    (J. Mol. Biol. 1994, 237, pp 125-156). The calculation of
    X-displacement, Y-displacement, Tip and Inclination is then exactly as
    described in SCHNAaP (J. Mol. Biol. 273, 668-680, i.e., 3DNA calculates
    a set of local helical parameters instead of linear global ones as given
    in SCHNAaP and NewHelix/FreeHelix.)

    To make the above point clear, let's use A1-A2-A3 triplet as an example.
    First, A1-A2 define a local helical axis and a set of local base-pair
    helical parameters are calculated. In 3DNA, these parameters are defined
    in a symmetric manner that bp A1:T1 and bp A2:T2 have exactly the same
    values except for a sign reversal for Y-displacement and Tip.
    Similarly, step A2-A3 define another set of local base-pair helical
    parameters. Thus bp A2:T2 has two sets of helical parameters associated
    with it depending on its context, i.e., either with bp A1:T1 or with bp
    A3:T3. Moreover, the local helical rise and helical twist are directly
    related to a dinucleotide step. These are the reasons that "Local
    base-pair helical parameters" as given in 3DNA refer to base-pair steps.

Please also refer to another post in this forum on "h-twist vs. twist" and the link therein.

HTH,

Xiang-Jun

1537

General discussions (Q&As) / Re: interpretation of output file

« on: September 13, 2008, 05:18:27 pm »

Again, as recommended clearly in the guideline, please provide a minimal, reproducible example to help others HELP YOU.

on your site there is only one instruction how to do find_pair, so I can run only this command.

As the author of "find_pair", I know clearly that it does not produce output on "deviation of the global helical axis from a regular linear axis".

I am certainly not in a position to answer the reviewer's questions to your paper. As a general rule, it is dangerous to use something you do not understand. You should clarify the issues beforehand instead of afterwards.

Overall, if you do not know how to use "fiber" to build fiber models,  not being able to find and follow the instructions in the 3DNA users' manual, clearly 3DNA is not the right tool for you. Curves is an excellent alternative and it is widely used to quantify DNA curvatures.

Good luck!

Xiang-Jun

1538

General discussions (Q&As) / Re: interpretation of output file

« on: September 10, 2008, 10:04:03 pm »

Well, I do not intend to be mean, but have you ever thought of answering the requests/questions in my reply to your initial post? Going through those questions would have clarified your understanding of the issues and helped other 3DNA users as well.

It might help to (re)read the post "Welcome message from Xiang-Jun Lu", and follow the link on "How To Ask Questions The Smart Way?"

Best regards,

Xiang-Jun

1539

General discussions (Q&As) / Re: interpretation of output file

« on: August 11, 2008, 09:56:53 pm »

Could you please tell me "Deviation from regular linear helix: 3.32 (0.37)".
Is it in angstroms or degrees? Is it a big bend or a mild one?

It is in angstroms. Based purely on the number, I would guess it is a mild one. As always, however, a number is just a number, and you should use a visualization tool (e.g., RasMol) to see/judge for yourself.

Is it possible to see "normal" DNA parameters to compare with the results from the output file? Which parameters other than the "Deviation from regular linear helix" indicate if the bend is big or mild?

Try use the "fiber" program to build regular B- and A-DNA models and analyze them. What would you get? Check and compare them with the examples directory (analysis/rebuilding) of A-, B- and nucleosome DNA structures adh006/bdl084/pd0001. It would be helpful that you summarize and post back a table of your findings.

In our structure, a protein interacts only with the half of the double-stranded B-DNA. How can I determine that only one half is bent?

3DNA works on a user-specified PDB file, which could be the whole structure or half of it.

HTH,

Xiang-Jun

1540

General discussions (Q&As) / Re: Rebuilding protonated RNAs with non-canonical basepairs

« on: July 30, 2008, 11:25:21 pm »

Sorry, I mis-formulated... I meant:
I can rebuild a U-U basepair, but analyzing does not work.

Well, it is not a problem of the 'analyze' program either. As a side note, how many such misunderstandings exist in literature?

Check FAQ #6 carefully: you should be able to get the answer. For your own understanding of the issue and to the benefit of other users, I am hoping that you would summarize your solution and post it back   . Otherwise, I will have no incentives on follow-up questions    

HTH,

Xiang-Jun

1541

General discussions (Q&As) / Re: calculate step parameters of a funny step

« on: July 23, 2008, 10:34:44 pm »

Read the FAQ #6. Specifically, check file 'misc_3dna.par': one of the geometric constraints is the angle between base normals, with a default of 65 degrees. The base pair 58 (58 G and 233 C) you referred to could be larger than it.

As suggested in previous forum posts, you can always manually edit the 'find_pair' output before feeding it into 'analyze' to get the parameters of arbitrary bps or steps you are interested in.

HTH,

Xiang-Jun

1542

General discussions (Q&As) / Re: Rebuilding protonated RNAs with non-canonical basepairs

« on: July 23, 2008, 10:27:15 pm »

There is nothing special about U-U pair in rebuilding. Could you please provide a minimal reproducible example? You could take advantage of the attachment functionality.

Xiang-Jun

1543

General discussions (Q&As) / Re: Multiple global helical axes in one structure, how to calc?

« on: June 25, 2008, 09:55:59 pm »

This question deserves as a FAQ now. Indeed, in the coming 3DNA Nature Protocols paper (anytime soon!), there is an example which involves quantifying the bending angle between two relatively straight helix regions in a DNA-RNA junction structure.

To answer your question now, yes, you need to analyze the two helical regions separately to get the helical axis vector for each, and then to calculate the angle between them. The "analyze" program always takes each structure fed to it as a whole, and outputs a ls-fitted global linear axis if the structure overall does not deviate significantly from a regular helix.

You might also want to play with Curves, which outputs a bending angle as part of its global analysis. As you know, this method is quite frequently mentioned in the literature.

HTH,

Xiang-Jun

1544

General discussions (Q&As) / Re: rotate_mol rotfile.dat

« on: June 11, 2008, 07:47:19 pm »

The format of the transformation matrix expected by 'rotate_mol' should be as follow:

Code: [Select]

   1  # x-, y-, z-axes row-rise
      0.0000      0.0000      0.0000
      0.3654     -0.9308      0.0000
     -0.1924     -0.0755     -0.9784
      0.9107      0.3575     -0.2067
It needs to be fed with option "-t=rotmat.dat": there is a typo in the help message, which has been fixed in v2.0. i.e.,

Code: [Select]

rotate_mol -t=rotmat.dat sample.pdb sample_rmat.pdbIn the coming 3DNA Nature Protocols paper, we have a concrete example of this functionality used to set an DNA-RNA junction structure with one helix region along x-axis, another (decomposed orthogonal component) along y-axis.

HTH,

Xiang-Jun

1545

General discussions (Q&As) / Re: What other software is available?

« on: May 06, 2008, 11:02:29 pm »

Thank for posting a 3DNA-related question in the forum. I am hoping more people would share their experiences to questions like this one.

Depending what you want to do, you might check NAB from David Case's group.

HTH,

Xiang-Jun

PS: It helps to provide a link when referring to a resource, e.g. model.it.

1546

General discussions (Q&As) / Re: Dealing with DNA-Protein complexes

« on: May 06, 2008, 10:40:31 pm »

Thanks for posting your step-by-step procedures: now the issue becomes quite clear.

But, as you said, both structures are using the 1st base-pair reference frame. In the case that the DNA structure is bent (like in 1BDT) using the 1st base-pair reference frame is an issue as the DNA structure is overlapping the protein structure. That is why I was asking if one could set another base-pair as reference frame.

So the answer to your question is yes: you can set with reference to any bp-frame within the structure just by setting the option such as -6 for the 6th bp. When given two numbers, such as -6,7, the middle frame of bp step 6-7 is used to the orient the structure.
[hr:2fhadkdj][/hr:2fhadkdj]
If you ran "frame_mol -h", you will see:

Code: [Select]

-n1[,n2] base-pair serial number(s) [no spaces around ,]and in the example session,

Code: [Select]

EXAMPLES
        To set the Dickerson-Drew dodecamer CGCGAATTCGCG duplex structure
        (bdl084.pdb) with its minor groove at the middle A6-T7 step facing
        the viewer:
            find_pair bdl084.pdb stdout | analyze
            frame_mol -m -6,7 ref_frames.dat bdl084.pdb bdl084_new.pdb
        Check Examples/Calladine_Drew/ subdirectory for more examples
As I mentioned in my response to your first initial post, in modeling DNA structure from a DNA-protein complex, you could divide the DNA structures into fragments: vary only those that need to change while keeping the other parts as in their original conformation (base + backbone). This would require some manual work, but is doable with the various components in 3DNA, at least in principle. Of course, this approach is purely geometric based, thus the mutated DNA structure could be distorted or has steric-clash with protein.  You need to use molecular graphics based editing tools to fine tune the structure, and/or minimize it using energy calculations.

HTH,

Xiang-Jun

1547

General discussions (Q&As) / Re: Dealing with DNA-Protein complexes

« on: May 03, 2008, 06:46:38 pm »

Is there any way to change that ? For example, I would like to use the middle pair reference frame.

So, what's your "middle pair reference frame"? How can 3DNA know it in advance?

The 1st reference frame convention of rebuild structure should not be a problem in practice. Just as you can use frame_mol to re-orient your original PDB structure, you can do the same with structures from rebuild. As long as the two structures are in a common frame, they are comparable. And that's point.

For your own exercise and to the benefit of other forum users, I hope you could go through a simple test case, and would be willing to post back the step-by-step procedures you follow. Up to this point, only a few of 3DNA users sum up and post back. Ideally, more 3DNA users would share their experiences.

HTH,

Xiang-Jun

1548

General discussions (Q&As) / Re: Dealing with DNA-Protein complexes

« on: April 30, 2008, 10:34:46 pm »

First I'd like to thank the author and maintainer of this very useful and powerful tool, it is one thing to create such a tool but it is way harder to add some real support to it.

Thanks for your kind words about our efforts on 3DNA -- it is indeed to my great gratification to see 3DNA turned into a useful tool in the field of nucleic acid structures . Over the years, it is largely to the credit of the world-wide 3DNA user community that is driving its further refinements.

[hr:1kbblvhn][/hr:1kbblvhn]
Modeling DNA structure in a DNA-protein complex (while keeping protein structure unchanged, or RNA tertiary structures model building more generally) is an area where 3DNA can play a more active role. In principle, one can/should keep the DNA backbone conformation as they are except for the part where the base-pair parameters are varied. In this regard, it is essential to set a common reference frame so that the fragments can be properly connected. This process can't be automated in a general sense, so some purpose-specific script would be necessary.

I understand your problem in a general sense. Without seeing a specific example, this is what I would suggest you try:

• Since a structure from rebuild is always set w.r.t. the 1st base-pair reference frame, you need to firstly set your original DNA-protein complex w.r.t. its first bp frame (using find_pair/frame_mol)
• Since structures with rebuild use fixed standard backbone conformation, it is not a surprise that backbone connection would be out of normal distance, and thus not connected. By default, a cut-off distance of 4.5 A between O3'(i)--P(i+1) is used (check misc_3dna.par).

HTH,

Xiang-Jun

1549

General discussions (Q&As) / Re: Small bug in find_pair

« on: April 14, 2008, 10:31:50 pm »

As always, I welcome bug reports related to 3DNA: the more, the merrier.

The "bug" in find_pair regarding 'w' vs 'W' chain ID issue in 1VSP is well expected, though this is the first case reported. In 3DNA, all the atom names, residue names, and chain ID characters are converted to upper case, so 'w' and 'W' is treated as the same. Thus, running find_pair against this entry gives expected result.

Making 3DNA case-sensitive with regard to chain ID is not a big deal. However, do you know of any PDB documentation showing that upper/lower case makes a difference for chain ID? How about residue name, or even the 4-letter PDB id? Did you also analyze this entry with other programs, e.g., Curves, FreeHelix? How do they deal with the chain ID issue? More generally, in all the PDB/NDB entries, how many of them have chain IDs that differ only in upper/lower cases as 'w' vs 'W' for 1VSP? In my understanding, this issue may well be due to the one letter chain ID limit in PDB format itself.

HTH,

Xiang-Jun

1550

General discussions (Q&As) / Re: Baselist.dat

« on: April 09, 2008, 10:01:47 pm »

Hi Pascal,

Glad you noticed the size limitation of base list. In v1.5, which is the one currently being used, the upper limit was set to 512, and that explains the differences with the two baselist.dat files your attached. Back in around year 2000, I thought 512 would be large enough when I processed all the NDB entries. Indeed you are the first user to uncover this limit!

As far as so many unrecognized base residues as in your case, it is normal and intended. In principle, one could auto-detect the uncommon residues to keep 3DNA running without bothering with the baselist.dat file. However, as a design guideline, I have put several "check-points" in the 3DNA pipeline so users have more control, and can know what's going on. You could write a utility program to facilitate your job. If you do go that way, you might want to contribute back your work so others can benefit from your effort.

In the coming release of 3DNA v2.0, I have an up to date baselist.dat file for all the NDB entries up to March 2008, and of course, the 512 upper limit is gone. Sorry, No specified release date yet!

HTH,

Xiang-Jun