Double helix groove width parameters from 3DNA

In the 3DNA output (from the analyze program) for a DNA/RNA duplex structure, there is a section on "Minor and major groove widths: direct P-P distances and refined P-P distances which take into account the directions of the sugar-phosphate backbones". The underlying algorithm is that of El Hassan and Calladine (1998). ``Two Distinct Modes of Protein-induced Bending in DNA.'' J. Mol. Biol., v282, pp331-343. Note that the P-P distances need to be subtracted by 5.8 Å to take account of the vdw radii of the phosphate groups (2.9 Å), and for comparisons with NewHelix/FreeHelix and Curves.

Using 3DNA fiber models #1 for A-DNA (calf thymus) and #4 for B-DNA (calf thymus), the groove widths are as follows:

                 Minor Groove        Major Groove
              P-P     Refined     P-P     Refined
-----------------------------------------------------
A-DNA (#1)      18.5      16.7      15.2      11.1
B-DNA (#4)      11.7      11.7      17.2      17.2
-----------------------------------------------------

From the above table, it is clearly that for A-DNA, the minor and major groove widths for the refined set are smaller than their corresponding non-refined counterparts (i.e., those based on direct P-P distances). For B-DNA, there are no changes between the two sets. It should be noted that in real structures (i.e., non-perfectly regular, as in X-ray crystal structures in the NDB/PDB), there are nearly always some differences between the refined vs. direct P-P distances. As a general rule, the refined set should be used.

One of the key structural differences between A- and B-DNA is their opposite groove dimensions: for B-DNA, the major groove width (~17 Å) is about 5 Å wider than the minor groove width (~12 Å); whereas for A-DNA, the major groove width (~11 Å) is narrower than the minor groove width (~17 Å) by a similar amount. Since the grooves provide binding sites, the difference between A- and B-DNA grooves has important implications in DNA (groove) recognitions by ligands or proteins.

In retrospect, I implemented the El Hassan and Calladine algorithm for calculating the groove widths mainly because of its simplicity: I can understand clearly how it works visually. The algorithm is described in a two-page appendix of the above cited paper. For those who are interest in DNA structures in general and how groove widths are defined in particular, I would strongly recommend them to read the appendix carefully and try to implement it: there is no substitute for first hand experience. For an idealized cases, as the above for fiber A- and B-DNA, the implementation should be straightforward. To be more realistic, an implementation should account for missing phosphate groups in some structures (for testing purpose, simply delete one P atom from a structure), for example.

As is obvious, 3DNA does not calculate groove depths. Over the years, I have actually been approached with requests/suggestions to provide such parameters to complement groove widths. However, for various reasons, none of the algorithms fits with 3DNA. As a general principle, I do not add new functionality to 3DNA simply for the seek of it. I must understand a new piece clearly in order to integrate it with the rest and to be able to respond concretely to possible questions from users.