Where does all the computer power go?

Read "All systems go" in The Economist, October 25th 2007.

A complex computer model of a heart, developed over more than 40 years, is being used to improve understanding of the ways in which drugs cause arrhythmias; around 40% of the compounds that drug companies test cause these arrhythmia, and if the side effect could be reduced or eliminated, drugs would be safer. Denis Noble of Oxford University, the creator of the beating-heart model, is now part of a consortium involving four drug firms — Roche, Novartis, GlaxoSmithKline and AstraZeneca — that is trying to unravel how new drugs may affect the heart.

Virtual drugs are introduced into the model and researchers monitor the changes they cause just as if the medicines were being applied to a real heart. The production of some proteins increases while others are throttled back; these changes affect the flow of blood and electrical activity. The drugs can then be tweaked in order to boost the beneficial effects and reduce the harmful ones.

Systems biology thus speeds up the drug-testing process. Malcolm Young is the head of a firm called e-Therapeutics, which is based in Newcastle upon Tyne. Using databases of tens of thousands of interactions between the components of a cell, his company claims to have developed the world's fastest drug-profiling system. In contrast to the two years it takes to assess the effects of a new compound using conventional research methods, Dr Young's approach takes an average of just two weeks. Moreover, the company has been looking at drugs known to have damaging side effects and has found that its method would have predicted them......

Ultimately, the aim is to build an entire virtual human for researchers to play with. But reductionism is still needed to get there. Human bodies are made of cells, and the best way to build a model body might be to construct a general-purpose virtual cell that can be reprogrammed into being any one of the 220 or so specialised sorts of cell of which the human body is composed. That, after all, is how real bodies develop. And a collaboration organised by the European Science Foundation is hoping to do just this, through what it calls the Blue Cell project.

Keeping track of the data needed to carry out systems biology on this scale will be a Herculean task, and may turn out to be the driver of future developments at the heavy-number-crunching end of the computer industry. Dr Noble is in negotiations with Fujitsu, a Japanese computer firm that is developing a machine capable of performing some ten thousand trillion calculations a second. That would make it the world's fastest computer, but it comes with a price tag to match—about a billion dollars. This is a little more than the $6m paid for that fictional bionic man, Steve Austin, even allowing for inflation. But it is only about a quarter of what the Human Genome Project cost. And this time, it might produce some answers that prove immediately useful.

Comment: I have been suggesting that connectivity to cell phones or to the Internet is but one kind of index of the digital divide. Those of us who use personal computers may feel that we understand the technology, but there is a huge gap between the power of a PC and that of a $10 billion computer used to model a human being at the level of sub-cellular chemical and genetic activity.

The firms that first master the technology described in this article should have a huge advantage in drug discovery. Even now, the discovery of really new pharmaceutical products is dominated by huge ethical pharmaceutical companies, which are the only ones that can afford the hundreds of millions of dollars per drug to prove safety, efficacy and effectiveness. These firms may devote a fifth of their sales income to research and development. They also dominate the sales of ethical pharmaceuticals globally, with huge incomes.

The big guys are likely to be the ones to appropriate super computers for drug discovery. They are likely to use the technological power from the models and super computers to dominate the value chain in pharmaceuticals. Firms in developing nations may participate in the value chain, perhaps packaging product and selling it in developing markets. They are unlikely to be able to appropriate a large portion of the benefits from the drugs, seeing the lions share going to big pharma giants.

Unless means can be found either

• to give enough purchasing power to the victims of diseases of poverty (or to surrogates acting on their behalf) to develop a market attractive to big pharma, or
• to subsidize the development of pharmaceutical products for the diseases of poverty

poor people in poor countries are not going to benefit very much from this exciting, but expensive information technology. This is an aspect of the digital divide that gets too little attention! JAD