Sir

During the years that followed the discovery of the DNA double-helix structure, whose fiftieth anniversary is currently being celebrated (see http://www.nature.com/nature/dna50), a wealth of fibre diffraction studies were carried out (reviewed in ref. 1).

However, the atomic details of the double-helical structure could not be confirmed until chemically pure oligonucleotides became available in the 1970s and allowed the study of single crystals.

An important breakthrough was the crystallization of an RNA mini-helix2. Oligonucleotide crystallography has provided valuable information and unexpected discoveries, yet there is plenty of scope for more.

The two first DNA model structures that were determined produced more surprise than confirmation. The first of these3,4 was published on 22 June 1978, obtained from d(ATAT) and showing only parts of a double helix in a rather complex crystal structure. A year later, the second structure5 uncovered left-handed DNA. It was only in 1980 that the right-handed double helix proposed in 1953 could be seen in atomic detail6.

Since then a wealth of crystallographic data have fully confirmed the double-helical nature of B-form DNA and the Watson–Crick pairing of the bases in this and most other forms of DNA.

In 1979 Z-DNA was discovered, followed by quadruplexes, Holliday junctions, Hoogsteen pairing and others, as detailed in the nucleic acid database (http://ndbserver.rutgers.edu). This database has been recently improved, and provides images of any available structure.

Some may think the field is mature, with the structures of more than 600 deoxyribonucleotides known. Yet these data are extremely limited; for example, only two structures have been determined with only AT pairs — most structures are rather CG-rich. Some sequences are easier to study, and so about 85% of crystallized B-form DNA models start with cytosine.