Dr. Paul W.K. Rothemund [1] and Dr. Eric Winfree [2] were awarded the 2006 Foresight Institute [3] Feynman Prizes at the nanoTX Conference [4] last week, at a special awards reception. (I earlier blogged about H. Ross Perot’s keynote address [5] at this conference.)
Rothemund was on-hand to receive the award, and I was fortunate to be able to attend his presentation on his and Winfree’s research. Rothemund delivered first a presentation on his work, and then he delivered a presentation on behalf of Winfree who could not attend.
Rothemund’s work is fantastic — he works upon Algorithmic Self-Assembly. He’s been able to program long strands of viral DNA such than when mixed in a suspension with other short DNA snippets (and heated slightly), the snippets or “staples” will bind to the long strand in particular order, causing it to fold back upon itself to form precise shapes. Rothemund has nicknamed what he does as “DNA Origami”, although the key concept is the ability to program the DNA to order itself into near two-dimensional, or even three-dimensional shapes. As proof of concept, Rothemund has programmed DNA to fold itself into words, stars, smiley faces, and other shapes.
[6]
Smiley composed of one long DNA strand
The staple snippets of DNA are not shown in this representation.
(Illustration copyright 2006 by Chris Silver Smith.)
In a typical reaction, Rothemund’s construction of these nanosmiles was accomplished in an estimated edition of billions of copies! As Winfree puts it, “Paul’s smiley faces constitute the most concentrated happiness ever experienced on earth.”
Other work done by Winfree’s research group has accomplished algorithmic assembly sequences which can count up by binary, and even be programmed to stop counting upwards after reaching a certain number. Winfree is apparently very interested in the possibility of Seed Technology — the concept that using the same sorts of processes whereby nature builds complex objects such as trees or elephants and simple objects such as crystalline lattices, we might be able to tap into processes which could automatically assemble objects for our everyday use. If one could drop a tiny “seed” — a programmed set of instructions — into a soup of elementary compounds and the seed might catalyze them into assembling themselves into complex objects.
The reason I find this work exciting and interesting (and the reason why I attended the nanoTX conference) was that I see a number of disciplines rapidly converging. Physics, chemistry, mechanical engineering, biology, and computer science are all coming together and the borders are blurring! Is the work in DNA computing programming or chemistry? It doesn’t matter.
In practical terms, Rothemund’s and Winfree’s work has proved that they can program DNA sequences, replicate those sequences on such a scale to produce massive quantities of fabric patches each of which are precisely arranged on a nanometer sized scale. While they did not invent the concept of instructing DNA to fold itself into precise new structures (this was invented by Professor Nadrian Seeman [7] of NYU and others), their work has built upon the concept in innovative ways, promising any number of potential applications.
One problem in nanotechnology is that no one has yet solved the puzzle of mass reproduction of nanoscale machinery. As other speakers at the conference mentioned, we have the ability to assemble exceedingly tiny objects, but the process is currently painstaking and slow, and a manufacturing method for mass production needs to be accomplished in order for nanotech to realize its potential. Algorithmic Self-Assembly would seem to be one of the prime approaches for enabling mass production.
Some have supposed that DNA could be used in the process for the building of nanoscale computer chips. It’s possible to attach nano tools at points along the length of DNA, so the algorthmic self-assembly process might be able to build those tools into tiny circuit boards.
There are cases where the DNA programming for complex self-assembly will have errors however, so at least one of the obstacles for realizing this potential may require more work in self-correction or mitigating code errors.
I’m also interested in their work from the standpoint of general programming theory. Algorithmic self-assembly reminds me heavily of The Artificial Game of Live — a fascinating game invented by the famous mathematician, John Conway, where in the playing field is a grid of “cells” which are living or dead. One designates which cells are live at the begining, and then “play” begins. The rules state that if a cell has a certain number of neighboring cells occupied, it will continue to live, and if there are too few or too many neighboring cells occupied, it will die. If an empty/dead cell has a certain number of neighboring cells occupied/live, it will spring to life. These rules are applied through each iteration, for as many iterations as desired. There are a number of fascinating patterns which can be set up initially to cause groupings of cells to appear to move across the field as cells die on one side and come to life on another. It’s like the computer model of bacterial cultures growing on a petri dish.
Many of us who started programming with the advent of affordable PCs actually cut our teeth by programming our computers to play out iterations of the Artificial Game of Life. I had so much fun with it when I set up my Commodore 64 to run Life iterations way back when.
The Game of Life is an example of Cellular Automata [8] or Automata Theory [9], which have been used to model complex systems, and have sometimes been taken as supportive evidence that development of Artificial Intelligence might be possible. Indeed, some lines of experimentation in AI have involved the attempt to create self-programming systems, and algorithms which could allow for “learning” through iterative testing and automatic self-adjustments.
Winfree and Rothemund are well aware that the DNA programming is similar to cellular automata, and they seem well-versed in the basics of computational theory which overlaps their research. A long DNA strand of instructions could be very like a Turing Machine.
A number of speakers throughout the conference suggested that key pieces of nanotech were likely within only about four to five years from being developed or solved. That’s not far off.
Nanotech and near-nano tech have already been affecting our lives, and how we interact with the internet. Microchips and thin-screen displays have allowed us to have affordable cellphone PDAs and have paved the way to users downloading of music and videos to handheld devices. Information is now becoming easily and rapidly distributable. Most people may not know yet what nanotech is, but they’re buying the iPod nano and wireless PDAs in droves. As the internet is evolving towards pulling in info from more and more sources, and associating that data in unforseen ways, I believe that progress in nanotech is likely to further accelerate the evolution. Tiny devices will allow more feeds of info into the net, and will also allow redistribution of net content back out to users and to the environment.
The barriers between our physical environment, information, and our selves are about to blur. You can see it already through the eyes of an electromicroscope: 21st century programmers have started issuing programming commands into the environment, and matter has rearranged itself into smiley faces! What could be more fundamental than this?