|Do quantum computers make us what we are? by Mark Buchanan
YOUR BODY is teeming with quantum computers. Marching along your DNA and floating around your cells, several hundred million of these minuscule devices are rearranging your molecules in super-efficient quantum fashion.
So, at any rate, says Apoorva Patel, a physicist at the Indian Institute of Science in Bangalore. According to Patel, these weird machines are essential to life. Every living thing from the greatest whale to the lowliest bacterium depends on an army of quantum computers to copy its DNA and put together its proteins.
To many biologists this seems like a bad joke. Received wisdom is that quantum physics, aside from a few minor details, has nothing to do with biology. Sure, it underlies the chemistry of all molecules, including biological ones, but the quantum weirdness is kept well out of sight.
Even physicists have reason to be scornful. For 15 years, they have struggled to build a quantum computer, a device that could exploit the peculiar properties of the quantum world to do calculations with a style and speed to put any ordinary computer to shame. Physicists generally concede that the task is so formidable that a practical quantum computer won't exist for decades.
So Patel's proposal, which he unveiled in an electronic preprint in February (http://xxx.lanl.gov/abs/quant-ph/0002037), is a radical one by anyone's standards. The forces of evolution, he claims, may have solved the problem of quantum computing several billion years ago. It's a startling idea--but if true, it could explain a puzzle at the core of biology.
Biologists have known for half a century that the sequence of bases along each strand of DNA encodes biological recipes for making proteins. Each base is one of four possible kinds--cytosine (C), guanine (G), adenine (A) and thymine (T). So there is a fundamental difference between the four-letter code of DNA and the strings of 0s and 1s in any computer, where there are only two alternatives. This is where the mystery begins: why four rather than just two?
A binary code ought to be better. Modern computers use only two characters so they can store information using very simple components. To store a 0 or 1, the transistors inside a microchip only need two states, "on" and "off". More characters in the code would demand more complicated and costly devices.
Binary logic also cuts down on mistakes. Imagine walking to a distant hilltop and then trying to transmit a message back to a friend. You might carry 26 flags, one for each letter, and try to spell out messages that way. But on a breezy day an E might look like an F, and a P like an R. You'd be better off using just two flags, one black, one white, and expressing letters as strings of the two. Then your friend would face nothing but simple black-white decisions, and you could be more confident in your communications. When it comes to handling information, computer scientists agree that binary is best.
So why doesn't biology use it? Several billion years ago, when the first self-replicating molecules were evolving, this simplest of all codes ought to have been the first to arise, and should have defeated other, more error-prone codes in the evolutionary race. Or might there be something mysteriously efficient about the number 4?
Patel thinks there is. To see why, we need to think in terms of computation. "Computation is nothing but the processing of information," he says, "so we can study what DNA does from the viewpoint of computer science."
A biological computation happens every time a cell divides: the data stored in one set of DNA molecules gets copied into another set. In a stretch of double-stranded DNA, bonds link the bases along one strand to those on the other, with every C bound to a complementary G, and every A to a T. Just before cell division, enzymes unzip the strands, exposing the bases to the cell's internal soup of raw materials. Another enzyme known as a DNA polymerase then marches along each of the two strands, triggering each base to pair up with a complementary base from the soup. Step by step, the polymerase copies the genetic information and creates two new double-stranded DNA molecules identical to the original.
But there's more to this than the simple copying of data. As Patel sees it, the soup of bases is like a disorganised database containing four kinds of entry. The polymerase's task is to find an entry of one particular kind. As the polymerase repeatedly searches for the right base in the alphabet soup, it is doing computations. And here lies the nub of Patel's idea: we would expect the polymerase to search in the best way possible. So what is the best of all possible ways to search a database?
In conventional computing, the best you can do is trial and error. To search for one kind of object in a jumble of N different kinds, you try one after another until you get lucky. This way you will find the right thing after an average of N attempts. For instance, it takes four tries on average to find a heart by cutting a shuffled pack of cards. This is just like the soup of bases, which would get shuffled by thermal motions after each attempt.
So molecular biologists assume that DNA polymerase works in the same way. Every so often, a base of some random kind wanders past the polymerase. It becomes attached to the growing chain if it happens to be the correct base, and wanders off again if it isn't. In that case, a polymerase would need to test an average of four bases before finding the right one. Normally, this is the best that can possibly be done. But, says Patel, it is possible to do better by exploiting one of the weirder consequences of quantum mechanics.
In an ordinary computer, a transistor can be either on or off, so a bit is always either 1 or 0. An alternative is to exploit quantum physics, and to store information using single quantum particles such as electrons. One might store bits in an electron's spin, for example, which can be either "up" or "down". The key is that the quantum world also allows other seemingly nonsensical possibilities: an electron's spin can be neither up nor down, but in a superposition of both. So a string of electrons can hold not just one distinct string of 0s and 1s, but every conceivable string all at once.
As a consequence, a computer handling information in quantum fashion could do parallel processing on an outrageous scale, testing many possibilities at the same time. In 1997 mathematician Lov Grover of the IBM Research Division showed that a quantum computer can search a database far faster than any classical device. It starts with a superposition of all the different items in a database, and alters this quantum state to amplify the desired item and make the others fade away. For a huge database, the time savings are huge, and even for smaller values of N the quantum procedure is faster.
Coincidentally, Patel and Grover were graduate students together at Caltech in the early 1980s. "We met again last year," says Grover, "through a mutual interest in quantum computing." To Patel, Grover's algorithm suggested an intriguing question: might biochemistry pull off a quantum computation?
Grover's mathematics gives an exact formula for the number of quantum attempts, Q, needed to find one specific element in a database of N things. It turns out that if N = 4, then Q = 1. In other words, a quantum computer can distinguish between four distinct possibilities with just one attempt.
Of course, it would also take a single quantum step to distinguish between two possibilities. But with a four-base code, DNA only needs to be half the length. So biology might have decided to use four bases instead of two so that replication of the molecule can happen twice as quickly.
For Patel's idea to work, the DNA polymerase would have to be able to manipulate the biochemical soup around it, watching over the base-pair bonding process to ensure that it occurs in a coherent, quantum-mechanical way. Each time the enzyme moves to a new base on the strand of DNA it is copying, it sets up a quantum superposition of the four bases that lie somewhere in its vicinity, with one ghostly component corresponding to each. According to quantum theory, such ghosts act like independent waves that move towards the exposed base on the DNA strand.
Next, Patel believes, the superposition of the four "incoming" waves starts to interact with the exposed base. This should alter the four waves in different ways, he says, making them interfere with one another in such a way that the ghosts for each incorrect base cancel out, while those for the correct base reinforce. At this point, the C-G-A-T superposition collapses, leaving the correct base bound to the DNA chain with a hydrogen bond. In other words, the enzyme should act like a sort of waveguide, ushering the component wave for the correct base into its proper resting place, while rejecting the others--carrying out Grover's quantum search in the process.
"The quantum search scheme he shows is very nice," says Grover, "although a few of the details are somewhat speculative. If true, it is another instance where nature first figured out how to do things better than us."
Perhaps the biggest "if" is whether the noisy environment within the cell would permit all this quantum business. The greatest obstacle to building a quantum computer in the lab is the need to isolate all its working parts from external disturbances, as almost any interference will destroy the fragile quantum dynamics. In all their attempts so far, physicists have tried to do this by cooling their apparatus down to near absolute zero. At the temperature inside a living cell, the enzyme and the four bases ought to suffer an annihilating storm of abuse, which should wipe out any possibility of quantum behaviour.
So DNA polymerase would somehow have to protect the environment around the growing DNA strand, permitting the quantum computation to go forward undisturbed. Patel points out that the configuration of electrons around atomic nuclei helps to shield some nuclear properties from their
environment. Nuclear spins remain in quantum superpositions for several seconds. He suggests that something similar happens in biochemistry.
No one knows whether DNA polymerases really have all these properties, and yet the idea may not be so ludicrous: quantum physics is not as foreign to biology as one might think. In photosynthesis, biology exploits quantum possibilities at a scale above that of single molecules. When a photon is absorbed by a photosynthesising cell, its energy excites an electron into a delocalised state spread out over tens of molecules.
Patel's proposal is more radical, in that it involves quantum superpositions of whole molecules. The more massive the object, the less obvious its quantum nature should be: lightweight electrons flaunt their quantum properties, while whole molecules are usually more coy. But some researchers have begun to suspect that all enzymes may depend on a quantum process involving protons--still 150 times lighter than bases, but 2000 times heavier than electrons. Last year, biochemist Judith Klinsman and colleagues from the University of California at Berkeley demonstrated that to speed up crucial cellular chemical reactions, some bacterial enzymes rely on the tunnelling of protons--a quantum process that allows a particle to pass through a barrier even if it hasn't got enough energy to climb over. What's more, they manage the feat even at room temperature.
Finding out whether DNA polymerases perform even more daring quantum tricks will require careful experiments. In the meantime, Patel is trying to see where else the quantum connection leads. Every protein in the human body is a string made from 20 different kinds of amino acid. Why 20? Here again, Patel thinks, the signs point to quantum computing.
To set the stage for the making of proteins, a strand of messenger RNA copies the genetic information from DNA and carries it out to a ribosome, one of the cell's protein manufacturing plants. The ribosome steps along the messenger RNA, and to each set of three base pairs attaches a tRNA, a stringy molecule with three base pairs at one end and an amino acid at the other. Once again, the ribosome faces a search: to build the right protein, it has to repeatedly find a tRNA corresponding to just one of the 20 kinds of amino acid in the soup.
The number 20 would seem to have little connection to anything. Patel points out, however, that this is just the right number to set up another super-efficient quantum search: for according to Grover's algorithm, a three-step quantum search can find an object in a database containing up to 20 kinds of entry. Like the number of bases, then, the number of amino acids seems to be just right if biology has set things up so that the protein manufacturing process is, in a quantum sense, as efficient as it can be.
"The numbers are certainly very provocative," says Grover. As Patel puts it, "This is the first time they have come out of an algorithm that performs the actual task accomplished by DNA." But do these numbers really point to quantum computers at the heart of life?
Evolutionary biologists are not convinced. "This field is rife with premature speculation," says Laurence Hurst of the University of Bath. "The history of the 20 amino acids problem has seen some of the most ingenious explanations, which at first looked even better than this one." They were all shot down in flames, he adds, when the biochemistry of the code was finally unravelled. With regard to the number of amino acids, Hurst points to one specific issue that Patel concedes is rather troubling: that the correspondence between tRNAs and amino acids isn't one-to-one. "There may be 20 amino acids," says Hurst, "but the same amino acid can get put onto different tRNAs, and the tRNA does the interacting. So it seems to me that there are more than 20 types to be found."
Even if Patel's idea won't stretch this far, it may still explain why there are four bases in the basic structure of DNA. "Apoorva generates a lot of ideas," says Grover, "and I think irrespective of how the biological and chemical aspects turn out, this one will make an impact." And after all, why wouldn't evolution exploit any quantum avenues open to it? If the cell spurns quantum tricks, wouldn't that need some explaining of its own?
Mark Buchanan writes about physics and other fields from the village of Notre Dame de Courson in northern France. He can be contacted at email@example.com
From New Scientist magazine, 15 April 2000.