Recent Posts

21

MD simulations / Re: Base pair number not staying constant in simulations

« Last post by xiangjun on May 21, 2025, 10:03:01 am »

Thanks for the detailed report. While @rkumar should be the primary contact for dnaMD related issues, I'd like to address the general question of why the base pair number not staying constant in simulations.

In 3DNA, the find_pair program is used to identify base pairs in an input structure. For MD trajectories, when each frame is processed with auto-detected base pairs, the numbers can fluctuate due to the dynamic nature of the system. The 3DNA suite includes the Ruby script x3dna_ensemble, and the beginning portion of the "x3dna_ensemble analyze -h" command is as follows. Basically, it requires a template base-pair input file, possibly generated with ‘find_pair’ and manually edited as necessary.

------------------------------------------------------------------------
Analyze a MODEL/ENDMDL delineated ensemble of NMR structures or MD
trajectories. All models must correspond to different conformations
of the same molecule. For the analysis of duplexes (default), a template
base-pair input file, generated with 'find_pair' and manually edited
as necessary, must be provided.

Usage:
        x3dna_ensemble analyze options
Examples:
        x3dna_ensemble analyze -b bpfile.dat -e sample_md0.pdb

In DSSR, the --nmr (or --md) option can be used with --pair-list-input to analyze MD trajectories with a customized set of base pairs of interest. See the DSSR User Manual for more details in Sections "3.13 The --nmr option" and "3.9 The --pair-list options".

Best regards,

Xiang-Jun

22

MD simulations / Base pair number not staying constant in simulations

« Last post by piia600 on May 21, 2025, 08:16:27 am »

Hi all,

I have managed to briefly analyze my simulations with dnaMD Python module. I managed to analyze the curvature of the unbound DNA but I ran into issues while analyzing the bound DNA.

The 3DNA calculations finish fine, but they identify different number of base pairs in different frames. There are 40 base pairs in the starting structure, but in some frames this drops to 35.

I can load the helical axis with step range defined:

complexdna.set_helical_axis('complex2/HelAxis_complex2_test.dat',step_range=True, step=[10,25])

But I can't continue the curvature analysis, because the command

complexdna.generate_smooth_axis(step_range=True, step=[10,25], smooth=500, spline=2, fill_point=6, cut_off_angle=180)

Leads to an error (which happens around the time when the base pair number goes very low):

Fitting spline curve on helical axis of frame 6000 out of 80000 frames
---------------------------------------------------------------------------
ValueError                                Traceback (most recent call last)
Cell In[24], line 2
      1 # Generate global axis by interpolation (smoothening)
----> 2 complexdna.generate_smooth_axis(step_range=True, step=[10,25], smooth=500, spline=2, fill_point=6, cut_off_angle=180)
      4 # Calculate curvature and tangent along global helical axis
      5 #freedna.calculate_curvature_tangent(store_tangent=True)

File ~/.local/lib/python3.9/site-packages/dnaMD/dnaMD.py:1322, in DNA.generate_smooth_axis(self, step_range, step, smooth, spline, fill_point, cut_off_angle)
   1318     sys.stdout.write("\rFitting spline curve on helical axis of frame %d out of %d frames" % (
   1319         frame_number, nframes))
   1320     sys.stdout.flush()
-> 1322 xsmooth, ysmooth, zsmooth, mask = fit_axis(bp_idx, frame_number, RawX, RawY, RawZ, smooth, spline, fill_point, cut_off_angle)
   1323 maskArray = mask
   1325 smoothX.append(xsmooth)

File ~/.local/lib/python3.9/site-packages/dnaMD/dnaMD.py:2621, in fit_axis(bp_idx, nframe, RawX, RawY, RawZ, smooth, spline, fill_point, cut_off_angle)
   2618 points = fill_point * len(orig_x)
   2620 nest = -1
-> 2621 tckp, u = splprep([orig_x, orig_y, orig_z], s=smooth, k=spline, nest=nest)
   2623 xnew, ynew, znew = splev(np.linspace(0, 1, points), tckp)
   2625 new_axis = np.array([xnew, ynew, znew]).T

File /usr/local/lib/python3.9/site-packages/scipy/interpolate/_fitpack_py.py:155, in splprep(x, w, u, ub, ue, k, task, s, t, full_output, nest, per, quiet)
     13 def splprep(x, w=None, u=None, ub=None, ue=None, k=3, task=0, s=None, t=None,
     14             full_output=0, nest=None, per=0, quiet=1):
     15     """
     16     Find the B-spline representation of an N-D curve.
     17
   (...)
    152
    153     """
--> 155     res = _impl.splprep(x, w, u, ub, ue, k, task, s, t, full_output, nest, per,
    156                         quiet)
    157     return res

File /usr/local/lib/python3.9/site-packages/scipy/interpolate/_fitpack_impl.py:175, in splprep(x, w, u, ub, ue, k, task, s, t, full_output, nest, per, quiet)
    173 wrk = _parcur_cache['wrk']
    174 iwrk = _parcur_cache['iwrk']
--> 175 t, c, o = _fitpack._parcur(ravel(transpose(x)), w, u, ub, ue, k,
    176                            task, ipar, s, t, nest, wrk, iwrk, per)
    177 _parcur_cache['u'] = o['u']
    178 _parcur_cache['ub'] = o['ub']

ValueError: Invalid inputs.

Also the 3DNA HelAxis*.dat files contained ---- for some base pairs which were not understood by dnaMD. I needed to replace them with 0.00 to get the file read.

Could you help me get the curvature analyzed also for the bound DNA? I only used the DNA atoms in the analyzed bound trajectory to get a comparison with the unbound DNA.

If needed, I can share some files for testing.

23

MD simulations / Re: Do I need gromacs to use dnaMD for simulations?

« Last post by piia600 on May 20, 2025, 03:20:40 am »

Thanks for the help!

Now I have managed to install both 3DNA and do_x3DNA for GROMACS. However, it seems do_x3DNA needs 'residuetypes.dat' file form GROMACS installation. It seems I was able to bypass this by downloading the file form github and putting it in the working folder. I thought you might want to put a note of this to the documentation.

Keep up the good work!

24

MD simulations / Re: Do I need gromacs to use dnaMD for simulations?

« Last post by rkumar on May 19, 2025, 12:31:02 am »

Do I need GROMACS version of dnaMD to analyze simulations or can I use the Python module of dnaMD without GROMACS for simulations?

No, dnaMD is a python package purely written in python. However, it requires output from do_x3dna,  which is now available for direct download and does not require GROMACS installation. You can use both do_x3dna and dnaMD without requiring GROMACS installation.

Thanks,

25

MD simulations / Re: Do I need gromacs to use dnaMD for simulations?

« Last post by xiangjun on May 16, 2025, 10:00:09 am »

Thanks for the x3DNA-DSSR software which works wonderfully for single PDBs.

Thanks for using DSSR and for posting your questions on the 3DNA Forum. As for the analysis of an ensemble, please see the DSSR manual , especially Section: "3.13 The --nmr option":

The DSSR --nmr (or --md) option automates the analysis of an ensemble, such as NMR structures in the PDB or snapshots from MD simulations. The input coordinates file must be in either the classic PDB format where each model is delineated by MODEL/ENDMDL tags, or the mmCIF format where each ATOM/HETATM record has an associated model number.
...
The --json option makes it easy to parse the output of multiple models pragmatically. In addition to NMR structures, trajectories from MD simulations can also be processed. Popular MD packages (AMBER, GROMACS, CHARMM, etc.) all have their own specialized binary formats for trajectories. By design, DSSR does not work on these binary files. They must be converted to the standard PDB or mmCIF format to be analyzed by DSSR. The combination of --nmr and --json makes DSSR directly accessible to the MD community.

I have some MD simulations I would like to analyze with dnaMD. I ran them with Amber but converted them to GROMACS .xtc + .pdb files for analysis.

Do I need GROMACS version of dnaMD to analyze simulations or can I use the Python module of dnaMD without GROMACS for simulations?

I am not a practitioner of MD simulations. Questions related to dnaMD are best answered by its developer: hopefully @rkumar will chime in. See the thread Update of do_x3dna package, which can be used with files generated by GROMACS.

PS. I am also lacking the link to download the 3DNA from the forum for some reason, my forum view is similar to unregistered users.

There have been too many spam registrations nowadays, so I must stay continuously vigilant to keep the Forum clean. You should now see the download link. Sorry for the inconvenience.

Best regards,

Xiang-Jun

26

MD simulations / Do I need gromacs to use dnaMD for simulations?

« Last post by piia600 on May 16, 2025, 05:50:16 am »

Hi all,

Thanks for the x3DNA-DSSR software which works wonderfully for single PDBs.

I have some MD simulations I would like to analyze with dnaMD. I ran them with Amber but converted them to GROMACS .xtc + .pdb files for analysis.

Do I need GROMACS version of dnaMD to analyze simulations or can I use the Python module of dnaMD without GROMACS for simulations?

PS. I am also lacking the link to download the 3DNA from the forum for some reason, my forum view is similar to unregistered users.

27

RNA structures (DSSR) / Re: Building G-quadruplexes

« Last post by xiangjun on May 05, 2025, 10:57:00 am »

Hi shr,

Following the discussion in the previous thread on "Rebuilding circular Z-DNA", as quoted below:

In addition to Z-DNA, I also work on other non-canonical DNA structures, particularly G-quadruplexes (G4s). I’m developing a method to construct ideal G-quadruplex models from sequence data by first arranging guanine bases into tetrads, then building in the backbone and loop regions.

I am glad to hear about your work on G-quadruplexes. Actually, I have recently revised the G4 module in DSSR, fixed existing bugs, and added new features. The g4.x3dna.org website has undergone a complete overhaul, enabling users to upload their own structures for dynamic G4 analysis. Additionally, the DSSR-G4DB database is being actively updated on a weekly basis as new PDB entries are added. See the four blog posts comparing DSSR with other related analysis tools on G-quadruplexes: ASC-G4, Webba da Silva nomenclature, ElTetrado and related tools, and CIIS-GQ.

Moveover, I am also interested in modeling G-quadruplexes, taking G-tetrad as the building block. There are quite a few other threads in DSSR I'd like to pursue further in the future. I'd certainly like to hear more about your approach on modeling G-quadruplex.

I dug into the code of DSSR for modeling G-quadruplexes, and found the following experimental (and undocumented) features. DSSR can model G-quadruplexes using G-tetrad as the building block, and allows users to specify the number of G-tetrads and twist angle (among other things). See below for two examples: one with 3 layers of G-tetrads and 0 degrees of twist angle, and other with 6 layers and twist=36, respectively.

Best regards,

Xiang-Jun

28

RNA structures (DSSR) / Re: Rebuilding circular Z-DNA

« Last post by xiangjun on May 05, 2025, 10:21:04 am »

Hi Di,

Are you still interested in the topic of modeling circular Z-DNA? I'm planning for a new release of DSSR (v2.5.3) which includes new features for modeling nucleic acid structures. It would be great to hear your feedback on how it works in your specific case.

I take user questions seriously as they provide valuable opportunities to enhance the software. Each piece of user feedback helps me think in ways I might not have considered otherwise. By analyzing feedback and integrating suggestions, DSSR becomes more robust and user-friendly. At the same time, I consistently adopt a systematic approach when introducing new features, ensuring they are thoroughly tested and reliable while addressing users' concerns.

Best regards,

Xiang-Jun

29

RNA structures (DSSR) / Re: Contour of dsDNA/dsRNA

« Last post by tengxj12345 on May 01, 2025, 12:28:19 pm »

Thanks for your detailed explanation. I will check it out.

30

RNA structures (DSSR) / Re: Contour of dsDNA/dsRNA

« Last post by xiangjun on April 30, 2025, 11:38:47 pm »

Thanks for your clarification. The two attachments are very helpful. Now I can use the following 3DNA commands to reproduce the results:

Code: Bash

1. find_pair coor_7972.pdb | analyze

The output file "coor_7972.out" has exactly the same parameter as the attached file "summary.txt".

Now back to your question:

it seems quite strange. The helix doesn't follow the structure of my DNA well. Is there anything wrong, or there are other output can better represent the contour?

The "strange" behavior you are observing is due to the sensitivity of helical parameters to local structural variations. There is nothing wrong as far as 3DNA goes. To verify this, you could try the following two things:

* Build a perfectly regular fiber RNA duplex model using the command below, and repeat your procedure. You should see a straight helix as expected. For example, see Figures 1 and 9 of the 2003 3DNA paper.

Code: Bash

1. fiber -seq=AAAAAAAAAA -rna fiber-RNA-A10.pdb
2. # or better yet, using DSSR v2.5.2
3. x3dna-dssr fiber --rna-duplex --seq=A10 -o=dssr-fiber-RNA-A10.pdb

* With the parameters from 3DNA analyze output (bp_step.par or bp_helical.par), you can run rebuild to generate a structure. The RMSD between the original structure and the rebuilt one should be close to 0 for base + C1' atoms. If you analyze the rebuilt structure, you should get virtually identical helical parameters as for the original structure. The analyze/rebuild reversibility is one of the core features of 3DNA and DSSR, originating from the SCHNAaP/SCHNArP pair of programs based the CEHS algorithm.

Hope this helps! Basically, what you are observing is the expected behavior of 3DNA.

That being said, for visualization purposes, one might want to smooth the local variations using Bezier curves or similar methods.

Best regards,

Xiang-Jun