Genesis Machines: The New Science of Biocomputing

  • Martyn Amos
Atlantic Books: 2006. 320 pp. £18.99 1843542242 | ISBN: 1-843-54224-2
Programmable computers have been developed that use DNA. Credit: A. PASIEKA/SPL

When visiting the Santa Fe Institute in autumn 1994, I was chatting with Chris Langton when Stuart Kauffman came running down the hall waving a manuscript and shouting unprintables. When we cornered him, he was slapping his forehead mumbling: “I could have done this, if I had only thought of it!” It turns out that 'this' was the first example of a DNA-based computation, carried out by Leonard Adleman of the University of Southern California, launching the field of DNA computation (Science 266, 1021–1024; 1994). Martyn Amos's book Genesis Machines looks back on the 12 years since this event, and speculates about the future of the increasingly intertwined fields of biology and computer science.

The computational feat reported in Adleman's seminal article was innocuous enough: examine a graph of seven nodes, and determine whether a one-way path exists that connects all the nodes once and only once (an example of the hamiltonian path problem). But the importance of this work did not lie in the sophistication of the problem, but in the fact that it showed that strands of DNA, mixed together in a vial, could be controlled such that their biochemistry could be viewed as a computation. And this is perhaps the central message that Amos tries to convey in his book: all physical systems can be viewed as performing computations; it is down to the skill of the investigator to make them perform useful ones.

Since 1994, DNA computation has advanced considerably, and programmable computers have been developed using DNA molecules only and with no moving parts. But for the reader to appreciate the story of how vials of DNA molecules could be coaxed to do something nature never intended, Amos has to reach far beyond the basic biochemistry of DNA, to the origins of computer science and complexity theory. Indeed, DNA computation was not the brainchild of biologists, but computer scientists who became fascinated by the 'digital' nature of DNA and its role in information processing. The roots go back to luminaries such as John von Neumann and Alan Turing, and Amos goes to considerable lengths to expose these foundations.

Perhaps one of the weaknesses of this book is a result of the unavoidable interdisciplinary nature of its subject. For the lay reader for whom this book is written, the concepts of Turing universality, the complexity of 'P versus NP', and the biochemistry of DNA are likely to be difficult subjects, and Amos's introductions include biographical sketches of their main protagonists. As a result, more than 100 pages go by before Adleman's pioneering experiment can be described in detail, and even then it is interrupted by a narrative of the trials, tribulations and ultimately triumph of Kary Mullis and his discovery of the polymerase chain reaction.

Nevertheless, the central discoveries in this rapidly evolving field are covered, and the inevitable critics of this endeavour are not overlooked. Amos does a good job of injecting some sobriety into the narrative, lest the reader believes that home computers running on DNA are just around the corner. Indeed, he points out right from the beginning that DNA computation is unlikely to replace its silicon-based forerunner, but rather be confined to special problems. What these will be is not at all clear, because the initial excitement about DNA's ability to tackle problems in the most difficult category, such as factoring a number into its prime components, quickly waned when it became clear that thousands of litres of DNA would be necessary to solve problems that are now just beyond the reach of conventional computers. But Amos insists that DNA computation is interesting in its own right because of the lessons it teaches us about computation in biology. The last two chapters are therefore devoted to endeavours only indirectly related to the quest for a DNA-based computer, namely DNA self-assembly and synthetic biology.

As Amos explains, the stability of DNA and its digital nature can be used to create — or should we say program — structures on a nanometre scale, a feat that is exceedingly difficult to achieve using non-biological material. Inspired by the pioneering work of crystallographer Ned Seeman at New York University, Erik Winfree and Paul Rothemund of the California Institute of Technology set out to show that DNA can be programmed to fold into arbitrary structures whose intricacy can only be revealed by atomic force microscopy.

Synthetic biology, on the other hand, concerns the creation of biological systems from scratch or from components that had other functions. Behind it lies the concept that we do not need to understand every single detail of a complex system in order to be satisfied with what we know. After all, very few people understand down to the level of the electronics the workings of my laptop computer. But this does not worry me because I know that someone knows how to make it, so there is sufficient knowledge to create another one just like it. The same can be said for biomolecular systems: we may never be able to comprehend every molecular interaction within a cell, but if we can build a system that behaves just like its natural counterpart, then perhaps we should be satisfied with that.

Amos's account of the different paths to synthetic biology covers most of the recent advances that have made the headlines. These range from pieces of DNA called 'biobricks' and the 'repressilator' to a way of coaxing yeast to produce the precursor for a malaria drug — a feat for which Jay Keasling, a biochemical engineer at the University of California, Berkeley, won Discover magazine's first 'scientist of the year' award.

But I have to take Amos to task for presenting the work of James Shapiro, a microbiologist at the University of Chicago, as if his investigations of the mechanisms of evolution somehow contradict darwinian evolution. Shapiro, who studies the many ways in which organisms can actively restructure their genetic material, believes that such processes should be more widely incorporated into evolutionary thinking. Although Darwin could not possibly have anticipated such processes, they follow the general darwinian paradigm and do not, in my opinion, necessitate a 'third way' somewhere between creationism and what Shapiro calls 'neo-darwinian orthodoxy'.

But apart from this quibble, this is an enjoyable book that could perhaps have profited from more illustrations to convey some of the computational and experimental methods. I recommend it to anyone interested in computation writ large who is not afraid to cross disciplinary boundaries that once seemed impassable.