Comparing DSSR bp-pars,DSSR simple-bpParams,& find_pair|cehs local bp-parameters

[chubetty]

Hi Xiang-Jun,
I ran DSSR and find_pair|cehs on the same pdb file and noticed that the base-pair parameter values for shear, stretch, stagger, buckle, propeller, and opening are all very different. Even in the same DSSR output file, the bp-pars (Line 6) have differences with the Simple-bpParams (Line 5). I have read through El Hassan and Callandine (1995), J. Mol. Biol, 251,648-664, but am still unclear about the mathematical basis for the parameters. Could you give a simple explanation as to why the values are different within the DSSR output file and between the DSSR and find_pair|cehs output files? How would you decide which values to use for structural analysis?
Thank you very much!
Betty

[xiangjun]

Hi Betty,

You've run into one of the subtleties on base-pair parameters between CEHS, 3DNA, and DSSR.

Simply put, the CHES scheme ("The assessment of the geometry of dinucleotide steps in double-helical DNA; a new local calculation scheme.") was implemented in SCHNAaP/SCHNArP ["Structure and conformation of helical nucleic acids: analysis program (SCHNAaP)", and "Structure and conformation of helical nucleic acids: rebuilding program (SCHNArP)"], which laid the foundation for the 3DNA analyze/rebuild programs. The differences lie in the adopted reference frames, which can lead to significant discrepancies among the different analysis programs, even for WC pairs (see  "Resolving the discrepancies among nucleic acid conformational analyses."). In particular, CEHS would give a Stretch of ~5.5 Å for a WC pair, roughly the distance between the centroids of the six-membered rings of A/G and T/U/C.

The 3DNA local parameters are based on the CEHS scheme but are calculated using the standard base reference frame. You can still get the authentic CEHS parameters from 3DNA 2.4 by using the cehs program as you noted. The numerical values would be the same as those from SCHNAaP. As of DSSR v1.9.0-2019mar26, you can also get the authentic CEHS parameters via the newly added --analyze-cehs option.

The 'simple' parameters were introduced for a more intuitive, qualitative description of non-WC pairs (e.g. Hoogsteen pair). See my blogpost "Details on the simple base-pair parameters" and linked therein.

I have read through El Hassan and Callandine (1995), J. Mol. Biol, 251,648-664, but am still unclear about the mathematical basis for the parameters.

This is understandable. It is worth noting the sentence "We thank Dr C. A. Hunter for his many useful suggestions and comments, and Xian Jung [sic] Lu for checking data. " in the Acknowledgements. There was actually a story behind this and why I developed the SCHHNaP/SCHNArP programs. Anyway, there is a long way between a text description of an algorithm and its robust implementation where many details must be properly taken care of. If you want to get to the bottom of CEHS, you could start with the cehs program distributed with 3DNA v2.4., or SCHNAaP.

HTH,

Xiang-Jun

[chubetty]

Hi Xiang-Jun,

Thanks for your response. I ran my PDB file on x3dna-dssr --analyze, x3dna-dssr --analyze-cehs, and find_pair|cehs and ended up getting different local base-pair and local step parameter values for all three.  I also noticed that the x3dna-dssr --analyze option is the only one that gives local base-pair helical parameters. The other two do not output those values. Is there a way to get local base-pair helical parameters for --analyze-cehs and find_pair|cehs? I am not sure how to discern which output values would be most accurate for my PDB structure. All the interactions (that are observed in the helical portion) in my structure are parallel-stranded noncanonical base pairs. Knowing this, what command/option would you recommend using to obtain the most accurate parameters for my structure?

Thank you!
Betty

[xiangjun]

Please provide a specific example to illustrate your points unambiguously.

Thanks,

Xiang-Jun

[chubetty]

Hi Xiang-Jun,

I have attached the three output files, in the order of x3dna-dssr --analyze, x3dna-dssr --analyze-cehs, find_pair|cehs. Again, the local base-pair parameter values are quite different among the three files. The local base-pair step parameters are similar, but slightly different between --analyze-cehs and find_pair|cehs. And only the --analyze option outputs the local base-pair helical parameters. And all of the base pairs are noncanonical parallel-stranded interactions.

Thank you,
Betty

[xiangjun]

Hi Betty,

Thanks for attaching three files which make your point clear (to me). It is indeed a confusing topic. Let me explain in more details below.

• x3dna-dssr --analyze -- this is an option recently introduced to DSSR v1.9.0-2019mar26. It is intended to replace the 3DNA 'analyze' program. So the output from this option should be identical to those from "find_pair some.pdb | analyze". It contain local base-pair, local step parametets, and local helical parameters, and a set of 'simple' parameters intended for a qualitative description of non-WC pairs. The local parameters, but not the 'simple' parameters can be used as input to the 3DNA "rebuild" program. The local parameters are based on the standard base reference frames.
• x3dna-dssr --analyze-cehs -- this option is intended to reproduce 'authentic' CEHS paramers as described in the 1995 El Hassan and Calladine JMB paper, and in SCHNAaP. The output does not contain helical parameters because these are notavailable in the original CEHS paper.
• find_pair|cehs -- this option is intended to reproduce 'authentic' CEHS paramers (1995 JMB), plus a set of "SCHNAaP global helical parameters" I introduced in the 1997 JMB SCHNAaP paper.

Now the last two options, x3dna-dssr --analyze-cehs and find_pair|cehs, should give the same 'authentic' CEHS parameters. In the current case, they do not: the reason is that your structure contains only non-WC pairs which affect the directionality of reference frames chosen, and thus the ending results. Try a DNA structure, e.g. 355D, you'd see that they indeed give the same results. The original CEHS scheme was intended to study the conformational variations of WC structures.

I'll look into the issue so that in future releases, the two options will give exactly the same numerical values, even for non-WC pairs. Technically, it is not a big deal to fix. In reality, this can be very confusions, as illustrated clearly in your case.

Also, note that the x3dna-dssr --analyze option is a new addition. I mentioned the --analyze-cehs variant simply because we touched the CEHS topic. I'd otherwise not explicitly document this 'feature' in the DSSR manual.

At the very beginning, I thought you were interested in implementing the 1995 JMB-described CEHS scheme yourself. For that purpose, short of contacting the original authors themselves, I pointed you to the SCHNAaP program and the "ches" programs in 3DNA v2.4 distribution. That's still my advice.

If you're interested in analyzing non-WC structures using DSSR, please start a new thread.

Best regards,

Xiang-Jun

[chubetty]

Hi Xiang-Jun,

Thank you for the clarification. Just to make sure I understood this properly, CEHS parameters should only be used to analyze conformational variations of WC structures. And since my structure has only parallel-stranded non-WC interactions, I should be using the base pair and step parameters outputted by x3dna-dssr --analyze?

Thanks,
Betty

[xiangjun]

Just to make sure I understood this properly, CEHS parameters should only be used to analyze conformational variations of WC structures.

Yes, the original CEHS parameters are designed for WC structures, as mentioned clearly in the abstract of the 1995 JMB paper.

"In this paper, we develop a new local Euler-angle-based scheme for assessing the internal kinematics or geometry of a general dinucleotide step in double-helical DNA. The geometry of a dinucleotide step is completely defined by: (1) the base-pair parameters that describe the relative position and orientation of one base with respect to the other in a standard Watson-Crick base-pair, and (2) the step parameters that describe the relative position and orientation of the two base-pairs. The key feature of our scheme is that it makes use of the concept of a mid-step reference frame..."

That said, the CEHS algorithm is generally applicable and it forms one of the cornerstones of 3DNA and DSSR, including my extension of the idea to a set of local helical parameters.

And since my structure has only parallel-stranded non-WC interactions, I should be using the base pair and step parameters outputted by x3dna-dssr --analyze?

Yes. For your parallel structure, the local base-pair parameters from DSSR and the 3DNA 'analyze' program make perfect sense. As shown below, the C+C and G+G symbols match the parallel nature of your structure. Moreover, the ~180 opening means the glycosidic bonds are in opposite directions. Combined with the --more option, you will also see the DSSR classification of the pairs.

Code: [Select]

Local base-pair parameters
     bp      Shear    Stretch   Stagger    Buckle  Propeller  Opening
   1 C+C     -5.49      0.38     -0.26     -3.43     -0.76    177.10
   2 C+C     -5.30      0.03     -0.39     -6.39    125.84    174.80
   3 C+C     -5.39      0.10      0.29     -1.40   -128.26    178.13
   4 A-A     -4.24      4.91      0.84      8.66     13.21    -84.79
   5 G+G      5.16      3.66      0.04      4.43      0.39    -92.09
   6 G+G     -5.15     -3.66     -0.04     -4.42     -0.44     92.09
   7 A-A      4.24      4.91      0.84     -8.67     13.21    -84.83
   8 C+C      5.39     -0.09     -0.29      1.40    129.30   -178.12
   9 C+C      5.30     -0.03      0.39      6.40   -125.45   -174.78
  10 C+C      5.49     -0.38      0.25      3.41      1.72   -177.11

The local step and local helical parameters, however, do not make 'intuitive' sense. For example, the Rise values are negative or zero for some steps in your structure. Nevertheless, these parameters (together with the local base-pair parameters) can still be used to 'rebuild' your structure which is accurate at the base-pair level (i.e., no backbone).

You may have a look at the 'simple' parameters. For your parallel structure, the output (from your original attachment) is as below, which should look more reasonable. These 'simple' parameters, however, are for qualitative description only. They cannot be used for the rebuilding purpose.

Code: [Select]

Simple step parameters based on consecutive C1'-C1' vectors
     bp      Shift     Slide      Rise      Tilt      Roll     Twist
   1 C+C      0.16      0.12      3.28      0.27      1.05     46.50
   2 C+C      0.09      0.02      3.08     -1.15      2.21    -21.39
   3 C+C      2.56      0.77      6.67     -1.49     -1.18     13.68
   4 A-A     -1.49      0.62      3.10      0.04      7.07     77.70
   5 G+G      0.00      2.38      3.22     -0.01     -7.85     15.13
   6 G+G      1.49      0.62      3.10     -0.03      7.07     77.69
   7 A-A     -2.56      0.77      6.67      1.48     -1.18     13.69
   8 C+C     -0.09      0.02      3.08      1.16      2.21    -21.40
   9 C+C     -0.16      0.12      3.28     -0.28      1.06     46.51
  10 C+C      ----      ----      ----      ----      ----      ----
Best regards,

Xiang-Jun

[xiangjun]

As a follow-up, I've updated DSSR to v1.9.1-2019apr06 and 3DNA to v2.4.3-2019apr06. Now the DSSR --analyze-cehs option and the 3DNA cehs program should give the same numerical values of the CEHS parameters, even for non-WC pairs and structures.

Best regards,

Xiang-Jun