Messages

1251

Feature requests / Re: chain continuation character in analyze

« on: August 06, 2012, 01:04:12 pm »

Please download the updated aug06 release of 3DNA v2.1beta. Now you can specify helix begin/continuation/end and isolated bp characters through option -chain_markers. For example,

Code: [Select]

find_pair 1egk.pdb stdout
    #  default as before
find_pair -chain_markers='o|x+' 1egk.pdb stdout
    #  with helix beginning character assigned to 'o'
The same option can also be applied to 'analyze'.

Also, as noted previously, the same strand P...P and C1'...C1' distances are now output in two decimals.

Have a try and report back how it goes.

Xiang-Jun

1252

Feature requests / Re: BI/BII issue

« on: August 04, 2012, 03:13:30 pm »

As noted in the -tor output file,

Code: [Select]

          e-z: epsilon - zeta
              BI:  e-z = [-160, +20]
              BII: e-z = [+20, +200]

The criteria are based on "Nucleic acid backbone parameters" at the Jena website. Do you have another definition?

Also, the BI/BII classification is purely backbone-based (epsilon and zeta torsion angles), not specific to "B-DNA" whose definition may require characterization of the base pair geometry.

Xiang-Jun

1253

Feature requests / Re: chain continuation character in analyze

« on: August 04, 2012, 03:01:52 pm »

In the analyze file, you insert the '-' character when a strand is not broken and 'x' when its broken.
Then a new chain starts. May be you could add a third '+' character for these residues ?

Could you provide an example?

Also, for the same strand P...P and C1'...C1' distances, could you add two decimals instead of one?

Done -- the distribution will be updated after clarification of the above point.

Xiang-Jun

1254

General discussions (Q&As) / Re: how to bend a big DNA soomthly?

« on: July 27, 2012, 10:29:11 am »

3DNA per se does not provide a prescription "to bend a big DNA smoothly". How to choose roll angles to fit a smooth curve is problem specific; 3DNA is mechanical and rigorous in that it constructs a structure corresponding to the parameters you fed into "rebuild", or (reversibly) it can "analyze" a given DNA structure to provide parameters that fully describe its base geometry.

As mentioned previously, and made clear in the 2008 3DNA Nature Protocols paper, the various prescriptions of roll-introduced DNA bending in Figure 3 are based on the classic work of Calladine and Drew. That protocol was intended to illustrate 3DNA's capability of building structures, in schematic representation, based on user-supplied parameters, not to show how to derive roll angles for any desired DNA bending.

That said, 3DNA is handy for constructing and visualizing a DNA structure in three dimensions to help verify if a roll prescription fulfills one's assumptions -- seeing is believing as well as understanding. For example, by noticing a bend structure in zigzag type, you immediately realized that your roll parameters (as sinusoidal) were not chosen correctly.

There are literature publications on how to fit a smooth ribbon to curved DNA. Pubmed or Google Scholar is your friend; it helps if you could share your findings.

HTH,

Xiang-Jun
 

1255

General discussions (Q&As) / Re: Analysis_assumptions!

« on: July 24, 2012, 02:41:05 pm »

The 3DNA program mutate_bases does exactly as its name suggests, i.e., mutate DNA/RNA bases, and from a pure geometric approach. It is up to the user (with any other desirable tools) to make sense of "structural and binding differences" between mutated and native structures.

Xiang-Jun

1256

General discussions (Q&As) / Re: mutate_bases error

« on: July 24, 2012, 02:35:49 pm »

mutate_bases 'c=Y, s=14, m=DC' 3DFV.pdb 3DFV_mut.pdb

Change comma to space, as below:

Code: [Select]

mutate_bases 'c=Y s=14 m=DC' 3DFV.pdb 3DFV_mut.pdb
Comma or semicolon is used to separate multiple mutations. I am updating command line help message to make this point clearer.

Xiang-Jun
 

1257

General discussions (Q&As) / Re: Question on hairpin-loop in PDB entry 3uzn

« on: July 20, 2012, 09:54:50 am »

• Well, with the excerpt from find_pair output, I can see what nucleotides (nts) you are referring to. As always, it's the details that count.

• I also thought you know what hairpin loop is, i don't know why, I'm sorry.

  I know what a hairpin loop is, but I certainly did not see what you were referring to in your previous post. Thanks for your wikipedia quota and link on "step-loop", however, I still miss your point as to how it is related to the helix (see attached figure) formed by nts 149-155 with 171-177 on chain A of PDB entry 3uzn.
• So at this point, I have no comment to your initial question: "Can you say something about it?"

• Oh, and I was using 3DNA ver 1.5.

  I am glad to know that 3DNA v1.5, which was compiled (nearly) a decade ago, is still in use. As noted in "Download instructions", that version is obsolete, and "no longer supported". So it's the time to upgrade to v2.1 (beta).

Xiang-Jun
PS. Note that I have once again split the posts on hairpin-loop from the original thread 'Helices and Isolates in output of find_pair' to make each one focused on a specific topic, and not too lengthy.

1258

General discussions (Q&As) / Re: Question on hairpin-loop in PDB entry 3uzn

« on: July 20, 2012, 08:14:11 am »

Could you provide us reproducible details how you got this "hairpin loop of length zero (between 155 and 156)"? Which version of 3DNA are you using? I did not see nucleotides 156-162 (chain A) at all in PDB entry 3uzn downloaded from the current RCSB website.

Xiang-Jun

1259

General discussions (Q&As) / Re: Meaning of base-pair id string e.g. 'U-**+-G'

« on: July 18, 2012, 11:13:04 am »

Thanks for your new question . I have split this post from the original thread 'Helices and Isolates in output of find_pair' to make each one focused on a specific topic, and not too lengthy.

I am sure I've answered this question before, maybe through priviate emails. I'll write a post on this issue on the 3DNA homepage soon [link added on 2012-07-30]. Here is the short answer to your question:

The general patten of a base-pair id string is M-XYZ-N for bases M and N. Only when XYZ equals --- would M and N be a possible Watson-Crick pair. If Z is '+', the two z-axes of bases are pointing the same direction and thus have a positive dot product. See my post "Hoogsteen and reverse Hoogsteen base pairs".

HTH,

Xiang-Jun

1260

General discussions (Q&As) / Re: Helices and Isolates in output of find_pair

« on: July 16, 2012, 09:47:15 am »

Thanks for providing a PDB list where find_pair cannot properly assign certain helices. It'd be more helpful if you could provide specific problematic helices, for at least some PDB entries, as you did for 1z58.

Best regards,

Xiang-Jun

1261

General discussions (Q&As) / Re: Helices and Isolates in output of find_pair

« on: July 15, 2012, 09:36:07 am »

Thanks for providing detailed info about a mis-assigned helix by find_pair. As shown in the attached image, in PDB entry 1z58, nucleotides 336-338 and 346-348 should indeed be assigned into a helix. While the helix assignment algorithm of find_pair works elegantly for the majority of cases, it is clearly not sophisticated enough to properly handle complicated structures such as 1z58 (the large ribosomal subunit from the eubacterium Deinococcus radiodurans): if you pay close attention to the output from find_pair, you will see warning messages in such cases.

I'm interested in refining 3DNA on those complicated structures, and your reported example is a concrete case to start with. Do you have more examples? The more varied and detailed cases I have, the easier to test find_pair against, and the more 3DNA can work for you in the end.

Xiang-Jun

1262

General discussions (Q&As) / Re: Helices and Isolates in output of find_pair

« on: July 13, 2012, 05:28:10 pm »

For model #5 of PDB entry 1aju, let's store its coordinates in file 1aju-m5.pdb. Run

Code: [Select]

find_pair 1aju-m5.pdb 1aju-m5.bpsyou get 1aju-m5.bps, with the following content:

1aju-m5.pdb
1aju-m5.out
    2         # duplex
   13         # number of base-pairs
    1    1    # explicit bp numbering/hetero atoms
    1   30  0 #    1 | ...5>A:..16_:[..G]G-----C[..C]:..46_:A<...5  1.26  0.86 43.33  8.60  0.14
    2   29  0 #    2 | ...5>A:..17_:[..G]G-----C[..C]:..45_:A<...5  1.05  0.14 21.15  8.79 -2.62
    3   28  0 #    3 | ...5>A:..18_:[..C]C-----G[..G]:..44_:A<...5  0.47  0.27 13.69  9.09 -3.30
    4   27  0 #    4 | ...5>A:..19_:[..C]C-----G[..G]:..43_:A<...5  0.55  0.43 20.29  9.06 -2.56
    5   26  0 #    5 | ...5>A:..20_:[..A]A-----U[..U]:..42_:A<...5  1.04  1.00 16.36  8.68 -1.15
    6   25  0 #    6 | ...5>A:..21_:[..G]G-----C[..C]:..41_:A<...5  1.01  0.44  9.82  8.78 -2.61
    7   24  0 #    7 | ...5>A:..22_:[..A]A-----U[..U]:..40_:A<...5  0.88  0.75 23.30  8.87 -1.45
   10   23  0 #    8 | ...5>A:..26_:[..G]G-----C[..C]:..39_:A<...5  1.16  0.44 20.37  8.79 -1.94
   11   22  0 #    9 | ...5>A:..27_:[..A]A-----U[..U]:..38_:A<...5  1.09  1.03 16.74  8.87 -1.00
   12   21  0 #   10 | ...5>A:..28_:[..G]G-----C[..C]:..37_:A<...5  0.94  0.83 22.27  8.99 -1.29
   13   20  9 #   11 x ...5>A:..29_:[..C]C-----G[..G]:..36_:A<...5  1.14  0.55 24.72  8.76 -1.53
   15   18  1 #   12 + ...5>A:..31_:[..U]U-**--G[..G]:..34_:A<...5  4.82  2.03 40.01  7.97  7.88
   16   17  1 #   13 + ...5>A:..32_:[..G]G-**+-G[..G]:..33_:A<...5  6.66  0.32 62.30  6.72  9.42
##### Base-pair criteria used:     4.00     0.00    15.00     2.50    65.00     4.50     7.50 [ O N]
##### 2 non-Watson-Crick base-pairs, and 3 helices (2 isolated bps)
##### Helix #1 (11): 1 - 11  ***broken O3'(i) to P(i+1) linkage***
##### Helix #2 (1): 12
##### Helix #3 (1): 13

The helix continues up to base pair (bp) "A:..29_:[..C]C-----G[..G]:..36_:A" (#11). The next two bps are isolated, i.e., not part of a continuous helix formed by bps from 1 to 11. Please see the attached blocview-image showing nucleotides 28 to 37, with green for G, cyan for U, red for A, and yellow for C.

Regarding your question,

When bp is represented with bigger number first (like 20-9) does it mean anything?

As shown above, the numerical values for bps (the left two columns) from find_pair are nucleotide sequential numbers as they appear in the input PDB file. What do you mean "does it mean anything?". As always, please be specific, using an example to illustrate your point.

HTH,

Xiang-Jun

1263

General discussions (Q&As) / Re: Helices and Isolates in output of find_pair

« on: July 13, 2012, 11:09:37 am »

Thanks for using 3DNA. In 'find_pair' output, '+' means isolated base-pair (bp), i.e., a bp not in a helical context. '|' means the bp is part of a helix, and 'x' means helix breaks at the bp.

It would help if you provide an example -- then our discussion would be more specific.

Xiang-Jun

1264

General discussions (Q&As) / Re: How to bend DNA in a protein-DNA complex

« on: July 12, 2012, 07:03:50 pm »

is there any way to incorporate protein information in the bp_step.par file such that the whole complex undergoes bend.

No. The rebuilding process in 3DNA is purely geometric, and it does not handle protein explicitly. If you have a DNA-protein complex to start with, bending DNA will most likely cause steric clashes.

Xiang-Jun

1265

General discussions (Q&As) / Re: How to bend DNA in a protein-DNA complex

« on: July 12, 2012, 05:39:38 pm »

Could you please make your point specific by using a concreate example? What do your mean "it is easy to make that bend DNA using 3DNA"?

Xiang-Jun

1266

General discussions (Q&As) / Re: create bases?

« on: July 12, 2012, 11:49:04 am »

The output file generated by analyze shows only 18 base pairs and does not include the first two TT in Y chain.

The first two TTs on chain Y are not paired to AAs in 3dfv: 'find_pair' identifies all 18 bps available and 'analyze' outputs only parameters for those 18 bps. Everything is as expected.

As for your first question, i.e., pairing the two TTs in chain Y with two AAs in chain Z, have you checked if the two ends are pseudo-continuous in crystal packing? If so, write the full coordinates in a PDB file, and then 3DNA 'find_pair/analyze' will work accordingly. If not, you need to model the pairing with certain assumption/approximations. 3DNA has no direct facility for the job, but may be tailored for the purpose. On the other hand, please have a check of Coot, NAB, PyMol etc tools which may fit the job better. Please report back how it goes, and I will consider adapting 3DNA for your requirement if no other practical approach is available.

Xiang-Jun

1267

General discussions (Q&As) / Re: A way to regenerate experimental coordinates from the reference coordinates?

« on: July 09, 2012, 01:11:45 pm »

Updated to 2012jul09 -- have a look and report back how it works.

Xiang-Jun

1268

General discussions (Q&As) / Re: A way to regenerate experimental coordinates from the reference coordinates?

« on: July 09, 2012, 11:26:31 am »

There are two aspects to your question:

• You want the original coordinates of the base pairs (bp) instead of those reoriented w.r.t. bp reference frames.
• You are concerned about the speed of transformations or extracting from PDB files.

Using Python or any other scripting language, you do not need to parse hundreds of PDBs, but only two: one is the the original PDB file, and the other is 'allpairs.pdb'. By scanning the later, you have a list of all bps, then you can extract one-by-one from the same original PDB file. Have a try and report back how it goes -- I do not think speed is an issue here.

That said, the most efficient way would be to add an option to 'find_pair', presumably -original_coordinate, so that instead of transforming to bp reference frame, the original atomic coordinates are written directly to 'allpairs.pdb'. I will update the 3DNA v2.1beta distribution soon, so stay tuned.

Xiang-Jun

1269

General discussions (Q&As) / Re: A way to regenerate experimental coordinates from the reference coordinates?

« on: July 07, 2012, 06:26:57 pm »

Thanks for using 3DNA and your kind words about it! Over the years, user feedback/encouragement has been the driving force to move the project forward.

As you noticed, 'allpairs.pdb' has each of its base-pair (bp) reoriented in the local bp reference frame. If you want the all-atom bp coordinates in the native PDB, you can write a simple Python script to extract them from the original PDB file -- certainly no "complex matrix transformation" required. To make the point clear, let's use '6tna' as an example.

Code: [Select]

find_pair -p 6tna.pdb 6tna.bps
    # This generates 'allpairs.pdb', where each bp is delineated by a MODEL/ENDMDL pair
head allpairs.pdb
        # The first 2 lines are as below:
    # MODEL        1
    # REMARK    Section #0001 ....>A:...1_:[..G]G-----C[..C]:..72_:A<....

For the first bp, you can then simply extract nucleotides G1 on chain A (A.G1) and A.C72 from '6tna.pdb'. Check $X3DNA/perl_scripts/pdb_frag for a rudimentary implementation in Perl. By looping through 'allpairs.pdb' and checking for pattern "REMARK    Section #", you can convert all the bps to their native PDB coordinates.

HTH,

Xiang-Jun

1270

General discussions (Q&As) / Re: which parameter shall I change to produce bend DNA

« on: July 05, 2012, 06:23:21 pm »

No idea if AMBER or CHARMM is parametrized for (strongly) bend DNA. Large DNA bending angle is normally found in protein-DNA complexes, e.g. IHF or CAP.

You will get more pertinent advice on this question in the AMBER or CHARMM mailing list.

Xiang-Jun

1271

General discussions (Q&As) / Re: which parameter shall I change to produce bend DNA

« on: July 05, 2012, 04:53:22 pm »

So, my question is what is the basis of choosing roll parameters for a specific DNA bending angle?

Good question. However, that's what 3DNA cannot answer directly. As is made clear in the manual you referred to and in the 2008 3DNA Nature Protocols (NP) paper, the formulae for roll-introduced DNA curvature were based on the work of Calladine & Drew.

3DNA's usefulness in this area is described in the 2008 NP paper:

The complicated three-dimensional nature of local bending makes it difficult, even for a seasoned scientist, to visualize how the variation in roll at different dinucleotide steps leads to global curvature. With 3DNA, one simply prepares a data file with any prescribed set of parameters and builds the structure to see what it looks like (Fig. 3). The matrix-based scheme adopted in the 3DNA analysis/rebuilding programs makes this a completely reversible and rigorous process.


HTH,

Xiang-Jun

1272

General discussions (Q&As) / Re: which parameter shall I change to produce bend DNA

« on: July 05, 2012, 03:51:01 pm »

Please explain figure 2 in the documentation.

Are you referring to recipe #2 of the 2008 3DNA Nature Protocols paper, "command-line script to create a 22 base-pair long schematic duplex structure with a 45° curvature per helical turn"? It was Figure 3, instead of 2.

All the (step-by-step) details about how to generate Roll-introduce DNA curvature are presented there. Just try to reproduce recipe #2, as another user recently went through. If you have any technical problems, please post back to the forum.

Xiang-Jun

1273

General discussions (Q&As) / Re: Problem with installing 3DNA on Mac OSX-Intel 10.4

« on: July 02, 2012, 04:54:19 pm »

Following my previous response, I asked for help from SBGrid on compiling 3DNA for Mac OS X Intel 10.4. Ben Eisenbraun told me the OS is on longer supported by SBGrid, and suggested two alternatives (see his message below).

I tried to compile a 3DNA universal binary for Mac OS X 10.4 ppc and intel without luck. Compiling a 10.4/Intel compatible binary on the 10.5/Intel machine went through, so I have created a tarball for you to test. Please let us know if it works.

Xiang-Jun

Hi Xiang-Jun,

> I recently received a report from a 3DNA user that has "Mac OSX
> version 10.4.11, with a 1.83 GHz Intel Core Duo." Neither of the
> compiled versions of 3DNA with "sbgrid-dev-flex, OS X 10.5,
> x86/x86_64" or "sbgrid-dev-cotterpin, OS X 10.4, PowerPC" works for
> this architecture.

We no longer have a OS X Intel 10.4 machine. I have two ideas for you
though:

- cross-compile a 10.4/Intel binary on the 10.4/PPC machine. In the gcc
  manpage:

       -arch arch
           Compile for the specified target architecture arch.  The allowable
           values are i386, ppc and ppc64.  Multiple options work, and direct
           the compiler to produce ``universal'' binaries including object
           code for each architecture specified with -arch.  This option only
           works if assembler and libraries are available for each architec-
           ture specified.  (APPLE ONLY)

- compile a 10.4/Intel compatible binary on the 10.5/Intel machine using:

export MACOSX_DEPLOYMENT_TARGET=10.4
export LDFLAGS="-mmacosx-version-min=10.4 -Wl,-headerpad_max_install_names"

Those are probably enough. You might need to get into sysroot arguments to
gcc if you are using an OS X framework.

-ben

--
| Ben Eisenbraun
| SBGrid Consortium                          | http://sbgrid.org       |
| Harvard Medical School                     | http://hms.harvard.edu  |

1274

General discussions (Q&As) / Re: Recipe 4 of 3DNA Nature paper: Questions

« on: July 01, 2012, 09:22:27 am »

The 1st line is a comment, while the next two lines define just the resolution of the Raster3D render-ed image. Check .r3d format for details.

In the $X3DNA/config/ directory, there are two files, my_header.r3d (default) and my_header_hres.r3d. Simply copy my_header_hres.r3d to my_header.r3d in the $X3DNA/config/ directory or to your working directory to use the high resolution settings. Alternatively, you can manually edit the .r3d file directly. Moreover, when a .r3d file is loaded into PyMol, the three lines have no influence at all.

Thanks for your posts -- I'm happy to see that you are able to fully verify our reported results.

HTH,

Xiang-Jun

PS: In "3DNA Nature Protocols Paper [vol.3, no.7 (2008), 1213-1227]" dated September 27, 2008, I wrote:

Importantly, in the NP_Recipes subdirectory distributed with 3DNA v2.0,  I have included all scripts, original data files and generated images, so that qualified researchers should be able to reproduce accurately our results without difficulty. Of course, I am more than willing and would be quick to address any reproducibility issues you may have. Repeatability is one of the basic requirements of a published scientific work, yet in this "high-throughput" / "big science" age reproducibility is sort of becoming a luxury.

In 3DNA v2.1, I've updated the scripts to ensure continuous reproducibility of our 2008 Nature Protocols paper.

1275

General discussions (Q&As) / Re: Recipe 4 of 3DNA Nature paper: Questions

« on: June 30, 2012, 07:08:26 am »

Hi Ignacio,

Thanks for your detailed questions regarding recipes #4 in our 2008 3DNA Nature Protocols paper. I am glad to know that you can reproduce recipes #1-3 and 5 without any problem. Check "What's New" item dated 2012-04-25 for switching from Perl to Ruby as the scripting language in 3DNA v2.1.

As mentioned at the very top of "script_h1x_h3y" for recipe #4:

# This script is a bit long, but it is actually quite simple.
# Try to understand exactly how it works, you would qualify as
# an expert 3DNA user!

Recipe #4 should be reproducible as the other ones. The best way to understand the procedures is to duplicate the whole directory to a new location, repeat and check each step as you move along.

How should I open the .r3d file to edit it? Is there a particular command or with a text editor like vi should be enough?

The .r3d file is in plain text, so you can view or edit it any way you like, e.g., using vi/emacs. Check Raster3D homepage for details about the .r3d format.

Is this blocview.r3d file the t.r3d file that I should've obtained?

The new Ruby version of 'blocview' generates the file 'blocview.r3d' instead of 't.r3d'. The final composite .r3d file is named '1egk_ok.r3d', which can be loaded directly into PyMOL. Note that the original Perl scripts are still available under $X3DNA/perl_scripts/.

All the scripts are distributed with 3DNA; reading through/between the lines is the most effective way to fully understand what's going on.

HTH,

Xiang-Jun