Paul Rothemund has written a paper that appears on the cover of this week's Nature (the link goes to a news story with an image of a DNA "smiley face", see the open access editor's summary, or the full paper (requires subscription)). He's come up with a way of folding a single DNA strand to form arbitrary two-dimensional shapes (much like origami experts fold a single sheet of paper into a multitude of designs).
Much work has been done on DNA nanotechnology, mainly inspired by the work of Ned Seeman. Rothemund's approach differs in that, rather than using many different "tiles" that self-assemble into a macro-scale single object, he folds a single section of well-sequenced viral DNA, using short "staple" strands to pin the whole complex together.
This work is wonderfully elegant, and is just one of several significant papers published by Rothemund. He first came to the attention of the DNA computing community at the first ever international workshop in 1995, when he described his scheme for a molecular Turing machine, and he's continued to produce work of outstanding quality.