segunda-feira, 23 de setembro de 2013


dnacomputer_Web.jpgAs a gadget to plug into a USB port, the “MinION” recently unveiled by Oxford Nanopore lacks the touch-me buy-me pizazz of Jonathan Ive’s designs. And since it’s a prototype that no one outside the company and a few partner organisations has yet been able to see in action, it is hard to say how well it actually works. But as an embodiment of technological cool it strikes me as pretty much beyond compare.

Inside the MinION is a little chip with 512 holes in it. Put some DNA into the MinION, and it will pull individual DNA molecules through those pores. DNA molecules carry genetic information in the form of four different chemical bases, like slightly different knots on a piece of string. As a DNA molecule goes through one of the MinION’s pores, the different knots on it are sensed electronically; the signals produced this way are processed inside the MinION and sent through the USB port to your computer, where the string of bases is reassembled as a genome sequence. How long are the pieces of string? The system can read individual strings tens of thousands of bases long—far longer than most sequencing technologies. A MinION should be able to read about a billion bases before its pores run out. That’s a third the length of a human genome. All in a device the size of a matchbox.

There’s no good way of putting a cost on the production of the first human genome sequence in the early 2000s, but the number people tend to quote is $3 billion. The technology in the MinION will apparently do it for well under $3,000. Getting a million times cheaper in ten years is quite a feat even by the standards of…well, by any standards at all. As a byword for head-spinning progress, we’re accustomed to thinking of Moore’s Law, which says (more or less) that the computing power available for a given price doubles every two years. But that gives you only a thousandfold improvement every 20 years. A millionfold in just ten really is something else.

The sheer sequencing power of the MinION by no means exhausts its claim to coolness. This summer sees the 100th anniversary of the birth of the great mathematician Alan Turing, and to understand anything of Turing’s work is to understand the fundamental importance of the ability to read what’s written on a piece of string. In the 1930s Turing tackled a fundamental problem in mathematics: whether it is possible to work out from first principles what things can be calculated, and what cannot. The tool he used was an imaginary machine that read numbers off a string and interpreted them as rules telling it how to write numbers back onto the same string. This simple idea allowed him to answer the question he was working on: no, it is not always possible to tell whether something is calculable. And the imaginary machine—the Turing machine—became the basis of computer science. A new book by George Dyson, “Turing’s Cathedral”, describes in detail how these insights were laboriously turned into primitive electronic hardware. But Dyson also sums up the big picture with almost gnomic elegance—information in computing “can take two forms: patterns in space that are transmitted across time, termed memory; or patterns in time that are transmitted across space, called code.” Both memories and actions can be represented by a simple string of numbers. Both cells and computers are Turing machines.

The way that the MinION reads the sequence data off single molecules letter by letter as they are ratcheted through its pores is a powerful evocation of Turing’s idea. It is not a Turing machine itself, it can only read, not write. But writing does play a crucial role in the creation of its all-important pores.

The pores are made of proteins inspired by nature but redesigned in computers. Those designs are then written on to DNA molecules. A DNA molecule is a memory, transmitting a pattern of information from one time to another. It is also a code, telling the machinery inside cells the sequence of steps that has to be taken to build a particular protein. If Oxford Nanopore could not write the DNA programs that tell bacteria to make the pores it needs, it could not build the devices that allow it to read DNA and thus decipher nature’s own programming.

The technology in the MinION may make Oxford Nanopore a lot of money. Then again, it may not. It is not fully proven as yet, and has a lot of competition, some of it making use of pores, some of it bringing the biological and the computational together in other ways. But integrating the biotechnology needed to make the pores with the computer technology that turns tiny electrical signals from those pores into DNA sequences on a computer screen is a tour de force. And it is also as cool an illustration as you could ask for of the deepest insight to flow from Turing’s work; that both biology and computer technology derive their power from nothing more, or less, miraculous than a string of numbers.

Oliver Morton is the briefings editor of The Economist and author of "Eating the Sun". He is a former editor of Wired (UK) and features editor of Nature

Illustration Gary Taxali


Nenhum comentário:

Postar um comentário