Different shear etc. 3 DNA and Curves+

[ellify]

I have the following question. I calculated the shear, buckle etc. for a DNA helix ( PDB 355d.pdb  same as the one in "Building a bridge between Curves+ and 3DNA") with 3DNA and with Curves+. I would have expected to get identical values for the shear, buckle roll etc. with both programs. However, that is not the case. e.g. shear for the first base pair using 3DNA is  0.28 and using Curves+ is 0.13, propeller -17.31 resp. -18.5

Why is there such a difference? Is there a different convention/parameters for the calculation in 3DNA compared to Curves+?

I produced the input for curves using  find_pair -c+ and I'm attaching all relevant files.

[xiangjun]

I have the following question. I calculated the shear, buckle etc. for a DNA dimer ( PDB 355d.pdb  same as the one in "Building a bridge between Curves+ and 3DNA") with 3DNA and with Curves+. I would have expected to get identical values for the shear, buckle roll etc. with both programs. However, that is not the case. e.g. shear for the first base pair using 3DNA is  0.28 and using Curves+ is 0.13, propeller -17.31 resp. -18.5

It is unrealistic to expect 3DNA and Curves+ to give identical values of base-pair parameters for a given structure. Even with the same Curves+ program, setting fit=.t. or not would give slightly different results. By adopting the standard base reference frame, 3DNA and Cuves+ give similar (but not identical) values for intra- or inter-base pair parameters. See the following blogposts:

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

Please also refer to the Web 3DNA 2.0 paper where further comparisons were made between 3DNA and Curves+.

Why is there such a difference? Is there a different convention/parameters for the calculation in 3DNA compared to Curves+?

As always, the devil is in the details. The differences between 3DNA and Curves+ come mainly from two aspects: (1) they implement the standard base reference frame differently; (2) they use different algorithms to calculate base-pair and step parameters. Since both programs are open source, you are welcome to dig into the technical details.

HTH,

Xiang-Jun