idealizing DNA geometry of existing model

[valpye]

Hi - I was wondering if there is a way to 'idealize' the geometry of a DNA model - in terms of pushing it towards B-form and C2'-endo sugar puckering? I have tried using 'find_pair *.pdb stdout | analyze stdin; x3dna_utils cp_std BDNA; rebuild -atomic bp_step.par *.pdb' and this provides a pdb file where the sugar pucker is corrected, and there are more sections which are B-form - however, the co-ordinates are shifted away from their current positions (and not all DNA sections are shifted in the same manner) and the chains are also renamed. I'm working on quite a large protein:DNA complex model to fit into an EM map and so would like to maintain the location and chain ID of the DNA co-ordinates and improve the geometry (to aid 'refinement' of this model against the EM map). What would be the best way to do this? Many thanks in advance!!

[xiangjun]

Hi,

I was wondering if there is a way to 'idealize' the geometry of a DNA model - in terms of pushing it towards B-form and C2'-endo sugar puckering?

The short answer is no. 3DNA does not 'idealize' an existing DNA model. However, there is a way around this 'deficiency'. See the FAQ entry "How do I build nucleic acid structures with sugar-phosphate backbone?" and my blog post titled "Restraint optimization of DNA backbone geometry using PHENIX".

I have tried using 'find_pair *.pdb stdout | analyze stdin; x3dna_utils cp_std BDNA; rebuild -atomic bp_step.par *.pdb' and this provides a pdb file where the sugar pucker is corrected, and there are more sections which are B-form - however, the co-ordinates are shifted away from their current positions (and not all DNA sections are shifted in the same manner) and the chains are also renamed.

The above procedure will generate a DNA model with exact base-pair geometry as in your original input PDB file, but only an approximate backbone: the sugars are all in C2'-endo as in the fiber model, and the O3'(i)--P(i+1) bond may well be out of the normal range. That's expected. You could use e.g. phenix (or other tools) to perform restraint optimization to get a more 'idealized' model.

I'm working on quite a large protein:DNA complex model to fit into an EM map and so would like to maintain the location and chain ID of the DNA co-ordinates and improve the geometry (to aid 'refinement' of this model against the EM map). What would be the best way to do this?

Two aspects here:

To maintain the location, you need to employ the utility program "frame_mol". See the example (frame_mol -h): [/li][/list]

Code: [Select]

EXAMPLES
        To set the Dickerson-Drew dodecamer CGCGAATTCGCG duplex structure
        (355d.pdb) with its minor groove at the middle A6-T7 step facing
        the viewer:
            find_pair 355d.pdb stdout | analyze stdin
            cp ref_frames.dat 355d_frames.dat
            frame_mol -m -6,7 355d_frames.dat 355d.pdb 355d_new.pdb
        and to rebuild a model with coordinates in the original frame:
            rebuild -atomic bp_step.par 355d_f1.pdb
            frame_mol -1 -r 355d_frames.dat 355d_f1.pdb 355d_raw.pdb
        Check examples/calladine_drew/ subdirectory for more examples
As for the chain ids of 3DNA generated models, they are fixed to A and B. You need to write a short script to change the ids as desired. This should not be that difficult to accomplish.

HTH,

Xiang-Jun