There are always exceptions to every "rule." However, for the most part, if the DNA is not perturbed by a drug or other molecule, it is generally accepted that octadeoxynucleotide duplexes will crystallize as A-form DNA, driven primarily by crystal packing forces, as you had described.

I think Pui Shing Ho is correct, but I'm not a crystallographer. I'd consult a specialist likeTom Brown (Southampton) on this one if I were you.

Thanks for your answer Dr. Ho and Dr. Baxter for suggestion. I have one more naive question. If I am understanding things incorrectly Please suggest me some references.

I am reading an article "Crystallization of DNA" by Timsit and Moras (http://www.sciencedirect.com/science/article/pii/007668799211022B) also. The paper suggests that  for Dickerson Drew dodecamer (DD) in P212121 family,  5'-CGC... and GCG-3' interacts at minor groove edge. But DD is completely B-DNA even after interactions at the end. So, if interactions cause change in conformation then end portions of DD in this family should be in A-form, but this is not the case. 

Look; please email this query to Professor Tom Brown;( [email protected]). He is the expert in this area. He is also a really nice guy and will point you in the right direction.

Thanks, Professor Baxter. I have communicated with Professor Brown as you suggested. 

Please read the Timsit and Moras paper more closely, and compare Figures 1 (the A-DNA packing arrangement) with Figure 3 (the Drew-Dickerson dodecamer packing). The A-DNA packing, which is seen in nearly all A-DNA structures, places the END of one duplex into the minor groove of an adjacent duplex (resulting in the duplexes being nearly perpendicular to each other).  The result is that the minor groove is pushed out and made more accessible while the major groove becomes more compressed and compact (both hallmarks of A-form DNA). Again, nearly all non-intercalated octameric DNAs do this and, consequently are forced into an A-conformation.  The Drew-Dickerson dodecamer has edge-edge interactions of the backbones towards the grooves (not the ENDS into the grooves). Thus, for the most part, octameric DNA duplexes are nearly all in the A-form because of crystal packing forces.  This was not well appreciated in the early days of DNA structure, but now with the very large database of such structures, it has become quite clear that there are crystal packing effects on many (not all) oligonucleotide structures.

I guess the only correction to your original statement is that the end-to-groove ineraction is to the minor and not the major groove of the A-DNA duplex.

Thanks for reply Dr. Ho. I am a computational chemistry student and was not much aware of Crystal packing based conformation dependence of DNA. your answer improved my understanding.