THE BODY MECHANIC

You may have already heard of big data, but you probably don’t know about LUDMIL ALEXANDROV, the researcher who’s identifying what causes cancer, one peta byte at a time.

THE ONLY ROAD MAP scientists have to help them cure cancer looks like one of those early renderings of the Amazon River – questionable tributaries and threatening blank spaces. The available information they have to help determine what causes the disease is even less instructive. To continue the map metaphor, it’s like a napkin sketch of Pietermaritzburg. Drawn in crayon. By a toddler.

To make sense of it, computational biologist Ludmil Alexandrov is sorting through tumour samples from 22 different countries at the University of California, San Diego, Supercomputer Center – on a computer so enormous it has a name: Comet. The idea is to let Comet crunch thousands of genetic samples until it zeroes in on the first molecular insults they have in common – mutations that can cause a normal cell to become cancerous. These so-called mutational signatures, and the biological or environmental factors that cause them, are what doctors and scientists have spent decades searching for: the causes of cancer.

So far, Alexandrov’s algorithm has found 90 mutational signatures. Many are created by known carcinogens, such as tobacco, UV light, and the breast cancer genes BRCA1 and BRCA2. ‘Probably 40 of them we understand well,’ he says. ‘But more than half, we just don’t know what they are.’

Alexandrov’s team feeds Comet reams of data about the samples, trying to find out why, for example, the rate of prostate cancer is 100 times higher in Scandinavia than in South America. ‘We try to associate the signatures we find with epidemiological characteristics, clinical characteristics, gene expression,’ Alexandrov says. Sometimes he gets a hit. A corn mould appeared in one analysis. A plant called Aristolochia, which is popular in Chinese medicine, popped up in another.

Here’s what you just read: A pervasive mould and a popular Chinese medicine both cause cancer, and one guy in San Diego with a supercomputer just proved it. Imagine if he could figure out what causes the other 50 or so signatures.

Unfortunately, determining the cause of an unknown mutational signature can take an enormous amount of time and money: Alexandrov’s latest project grant, to use the computer for an analysis of just 5 000 patients, clocked in at $30 million (±R400 million), and even with that much, his team couldn’t evaluate every variable they wanted. It’s disappointing that the work is so expensive, Alexandrov says, because scientists with less money might have smart ideas for what to test, and finding even a single new cause of cancer could drastically reduce deaths.

For example: When another team expanded Alexandrov’s work on Aristolochia, they found that the plant can be considered responsible for 80 per cent of liver cancer in Taiwan. ‘If you find something that causes 80 per cent of cancers, you can think about preventing 80 per cent of cancers,’ says Alexandrov.

That statement may sound tautological on the surface. But read it again. Because wow.

I want to see how this interacts with our genetic material. Are some people immune?

– Ludmil Alexandrov, computational biologist, UC San Diego