Failed Downloading MD Ruby Scripts of 3DNA

[Moi Zhang]

Hello!

I am new to 3DNA and would like to use 3DNA for my MD trajectory analysis. So far, I have installed the standard 3DNA v2.4 and know that MD Ruby scripts and do_x3dna are two options for me.

However, I got a problem on entering  http://3dna.rutgers.edu:8080/data/x3dna_md_v0.1.tar.gz website. The browser said "this site cannot be reached". I further tested 3dna.rutgers.edu website and it still failed. Is it a problem of the server?

I am really expecting to try 3DNA and looking forward to the reply:)

Best regards,
Moi

[xiangjun]

Hi Moi,

I am new to 3DNA and would like to use 3DNA for my MD trajectory analysis. So far, I have installed the standard 3DNA v2.4 and know that MD Ruby scripts and do_x3dna are two options for me.

Thanks for your interest in using 3DNA and for posting on the 3DNA Forum. The Ruby scripts distributed with 3DNA v2.4 may be appropriate for the analysis of an NMR ensemble in the PDB or short snapshot of MD trajectory. Check $X3DNA/bin/x3dna_ensemble script and $X3DNA/examples/ensemble.

"do_x3dna" is a third-party tool based on 3DNA. As far as I know, the "do_x3dna" project is no longer supported. Search the 3DNA Forum may give you some info.

However, I got a problem on entering  http://3dna.rutgers.edu:8080/data/x3dna_md_v0.1.tar.gz website. The browser said "this site cannot be reached". I further tested 3dna.rutgers.edu website and it still failed. Is it a problem of the server?

The 3dna.rutgers.edu server is hosted by Rutgers University. You may get any information on it from Dr. Wilma Olson. I have nothing more to offer on this matter.

You may try DSSR with the --nmr and --json options for MD analysis. As noted recently in the announcement post "DSSR 2.0 is licensed by Columbia University", "DSSR 2.0 supersedes 3DNA 2.4, which is still maintained but no additional features other than bug fixes are scheduled."

Best regards,

Xiang-Jun

[Moi Zhang]

Hi Xiang-Jun,

Thank you for your detailed reply! Considering the microsecond-level AMBER trajectories and the situation of 3DNA, I finally asked the license of DSSR. I am still waiting for their response:)

Hopefully, DSSR can solve my problem, I would like to try it and then have feedback here.

Thanks again!

Best regards,
Moi

[xiangjun]

Hi Moi,

Considering the microsecond-level AMBER trajectories and the situation of 3DNA, I finally asked the license of DSSR. I am still waiting for their response:)

Yes, DSSR 2.0 is the way to go. See the overview PDF.

Please let me know if you still do not hear back from CTV by early next week.

Hopefully, DSSR can solve my problem, I would like to try it and then have feedback here.

I am curious to know why do you want to use 3DNA/DSSR for the analysis of MD trajectories? Aren't there already dedicated tools (e.g., AMBER: CPPTRAJ) that *should* do the job, conveniently? In other words, what are still missing in those ready-to-use tools? I may be interested in extending MD analysis features in DSSR 2.0, ONLY IF justified by real-world applications.

Best regards,

Xiang-Jun

[Moi Zhang]

Hi Xiang-jun,

Thank you for the detailed information! I got the license from the university last week, and then tried to install and run DSSR. Unluckily, I noticed from the manual that DSSR only support PDB and mmCIF format... while my MD trajectories are millisecond-level netCDF file... If I convert the trajectory to PDB files, it would be over 30 GB... So I cannot continue with it.

As for my motivation to use 3DNA, I started working on Holliday Junction (HJ, a kind of DNA structure with four strands forming two helices, and two of the strands are shared by the two helices) recently by computational methods. Usually, the HJ conformation can be described by directions of helices. I hope that I can use the definition in 3DNA to describe the helix vectors so that I can see how those confirmations were sampled during the simulation (I failed to deal with it with cpptraj). Also, I want to monitor the base-pair H-bond along the helices during the simulation. Although cpptraj can solve it but this work is quite tedious:(

I also read some literatures and 3DNA and Curves+ are the only methods mentioned by those authors... Frankily speaking, I don't think Curves+ has a decent manual to explain its usage... I hope I am wrong

Best regards,
Zhengyue

[xiangjun]

Hi Zhengyue,

Thank you for the detailed information! I got the license from the university last week, and then tried to install and run DSSR.

Glad to hear that you have been able to download and run DSSR successfully.

Unluckily, I noticed from the manual that DSSR only support PDB and mmCIF format... while my MD trajectories are millisecond-level netCDF file...while my MD trajectories are millisecond-level netCDF file... If I convert the trajectory to PDB files, it would be over 30 GB... So I cannot continue with it.

As noted explicitly in the DSSR manual, it is a deliberate decision to support only the standard .pdb and .cif formats. I am not a practitioner of MD simulations. Presumably, any decent MD packages should have a way to convert its proprietary binary format to one of the two standard ones. Large converted file size (30GB in your case) is indeed a technical issue. DSSR may not be a straightforward solution to your case yet. If you find a solution elsewhere OR come up with one of your own, please post back so other viewers of the thread can benefit from your experience.

As for my motivation to use 3DNA, I started working on Holliday Junction (HJ, a kind of DNA structure with four strands forming two helices, and two of the strands are shared by the two helices) recently by computational methods. Usually, the HJ conformation can be described by directions of helices. I hope that I can use the definition in 3DNA to describe the helix vectors so that I can see how those confirmations were sampled during the simulation (I failed to deal with it with cpptraj). Also, I want to monitor the base-pair H-bond along the helices during the simulation. Although cpptraj can solve it but this work is quite tedious:(

So it seems that 3DNA/DSSR  does have something unique to offer. Could you provide a small, typical example file to illustrate unambiguously what you want to achieve using 3DNA manually, and how it it solved with cpptraj?

To me, file size is only a technical issue. If DSSR 2.0 can indeed offer features not (easily) available elsewhere, save MD practitioners large amount of time, THEN I'd like to come up with a practical SOLUTION. I need a compelling case to be made. Otherwise, what's the point, why bother?

I also read some literatures and 3DNA and Curves+ are the only methods mentioned by those authors...(

3DNA and Curves+ have complementary features. Specifically, Curves+ has more parameters for quantifying groove dimensions and helix curvatures, and better/integrated support for MD simulations than 3DNA. You may ask "those authors" how 3DNA and Curves+ were used in their cases. See my blogposts:

• Building a bridge between Curves+ and 3DNA
• Analyzing DNA/RNA structures with Curves+ and 3DNA
• Curves+ vs 3DNA

Best regards,

Xiang-Jun

[Moi Zhang]

Hi Xiang-jun,

Sorry for the late post. I was doing some other things and also understanding the usage of DSSR last week.

I am glad that I found out a way to apply DSSR on my trajectory without taking too much space.
The protocol is:
1. Divide the whole trajectory into let's say 1000 parts, and I would like to have 100 frames in one trajectory file. In my case, the simulation already divided the entire trajectory into small sections.
2. Use cpptraj to convert a single trajectory file into PDB file. Here, I only focus on DNA so I strip water and ions.
3. Run DSSR like

Code: [Select]

x3dna-dssr -i=production500.pdb -o=pro500.json --more --nmr --json. Like what you suggested, I use json output here.
4. Following the manual, I used jq tool to extract information from the json file I got on step 3. E.g.

Code: [Select]

jq '.models | .[] | .parameters.stems | .[] | {stem: .index, helical_direction: .helical_axis}’ <json file>.
5. Repeat the steps 2 to 4, and then I can retrieve helical axis information of the whole trajectory. Then I just need to concatenate them. Just make sure I only keep the useful files during the iteration so that I would not occupy too much space.

Meanwhile, I notice that, when numbering the stems, I guess the program number the stems by their residue number (from 5' to 3'), namely, the helix with smaller resid at 5' side will be numbered first... Is this correct?

Thanks so much:)

Best regards,
Moi

[xiangjun]

Hi Moi,

Thanks for sharing your way of applying DSSR to the analysis of MD trajectories. The protocol you described is exactly a DSSR user-case I have in mind. DSSR is not targeted specially for MD simulations, for sure. Yet, DSSR fits pragmatically in most situations involving DNA/RNA structural bioinformatics, by design. The MD community will realize the simplicity and applicability of DSSR: it is just a timing issue (when, but not if).

Meanwhile, I notice that, when numbering the stems, I guess the program number the stems by their residue number (from 5' to 3'), namely, the helix with smaller resid at 5' side will be numbered first... Is this correct?

Yes. Try DSSR on 1ehz or other examples you are sure of to check this out.

Xiang-Jun