find_pairs output

[elpierco]

Dr. Lu,

  Thanks for the quick response.  Is it possible to force find_pairs from reverseing the direction of base pairs.  For example:

 44  251  0 #   44 | -:..44_:[THY]T-----A[ADE]:.104_:-  0.98  0.91 27.73  8.86  1.31
   45  250  0 #   45 | -:..45_:[GUA]G-----C[CYT]:.103_:-  0.35  0.33 11.70  9.05 -0.49
   46  248  0 #   46 | -:..46_:[THY]T-*---T[THY]:.101_:-  1.48  1.02  9.88  9.32  3.51
   47  249  9 #   47 x -:..47_:[ADE]A-*---A[ADE]:.102_:-  6.37  4.93 16.04 10.30 16.23
   48  247  0 #   48 | -:..48_:[ADE]A-----T[THY]:.100_:-  0.91  0.88 36.14  9.06  1.17
   49  246  0 #   49 | -:..49_:[ADE]A-----T[THY]:..99_:-  0.91  0.88 51.76  8.99  1.18
   50  245  0 #   50 | -:..50_:[THY]T-----A[ADE]:..98_:-  1.10  1.01  8.23  9.02  1.63


I have been looking at the structure with VMD and it does look like a mess around these residue locations.  Just wondering if you have run into something similar.  I am using the following base-pair criteria
7.00 15.00  6.50 75.00  4.50  8.50

This is just one snapshot in the simulation and the other snapshots leading up to this and afterwards dont have this problem.  Out of about 250,000 snapshots approximately 6000 show a similar problem.  Thanks, Levi

[xiangjun]

I am well aware of the issue you experienced here. As documented in the FAQ section of the 3DNA home page, the geometry-based algorithm implemented in "find_pair" works well for what it has been designed for. This is one of the key utility programs in 3DNA that has made it possible to analyze nucleic acid related structure automatically. BTW, it is the method used in the NDB as well.

In your case, "find_pair" does find the pair, only not the ones you would expect intuitively. As you mentioned "... looking at the structure with VMD and it does look like a mess around these residue locations." As currently implemented in "find_pair", for example, T46 is thought to better match with T101 than with A102. If you could provide me with some typical example structures (via email), I will try to see if I could improve the situation.

In your case, if the 250,000 snapshots correspond to the same structure (thus with the same bps), you could simply start with one that works and modify it for all the others (only the top two lines). Have a look at http://rutchem.rutgers.edu/~xiangjun/3DNA/manual.html for a description of the format from "find_pair". I think you do not have to run "find_pair" for each of the 250,000 snapshots.

Have a try and let me know how it goes. Please also send me some sample structures with this problem via email.

Xiang-Jun