A patient-profiling study that may explain why symptoms of malaria vary widely between individuals began as a simple chat on board an aeroplane. Computational biologists Aviv Regev and Jill Mesirov, now colleagues at the Broad Institute, of the Massachusetts Institute of Technology and Harvard University in Cambridge, happened to be returning from a meeting together, before Regev formally joined the institute. The two struck up a conversation that turned into the beginnings of a fruitful collaboration before the aeroplane even touched down.

The malaria parasite Plasmodium falciparum has a notoriously complex life cycle. Patient reactions to infection range from asymptomatic to mild flu-like symptoms to severe fever, inflammation, coma and death, but so far nobody has managed to explain the biological basis of this variation.

Regev's interest lies with complex molecular networks, which she has generally studied in yeast, and Mesirov specializes in molecular pattern recognition, working mostly in cancer. But the two came together over the work of a colleague, Harvard University infectious-disease specialist Johanna Daily, who was working with patients with malaria in Senegal.

Lab-based studies of the gene expression of isolated malaria parasites had found no significant changes under different conditions. But Daily suspected that plasmodia growing in patients' blood cells might show more variation, which, in turn, could help to account for the different disease states. Two years ago, she approached Mesirov with the problem and some preliminary transcription data, collected from parasites infecting her Senegalese patients.

Using techniques she had developed to study the profiles of cancer cells, Mesirov set about trying to extract patterns from Daily's data, and found that the P. falciparum samples fell into three distinct groups. But it wasn't until her airborne chat with Regev that a possible means to understand the clusters' biological basis became clear.

From the literature, Regev had already collected gene-expression patterns for yeast growing under different environmental conditions — normal, starvation and environmental stress. “I wondered whether we could try to project those signature expression patterns from yeast to Plasmodium,” she says, “as Jill had already done from mouse to human cancer.”

Because Mesirov had already developed the relevant computational tools, Regev adds, “It took about a day to go from 'oh, we have an idea' to 'oh, we have a result'.” They established that the three groups that Daily's P. falciparum samples fitted into corresponded to the physiological states that Regev had deduced from yeast (see page 1091). Now, the team is pushing to connect the various physiological states to disease outcomes — and, they hope, to reveal new treatment targets.

Regev and Mesirov hope to join Daily in Senegal next year for a taste of the field-work. In the meantime, says Regev, they don't mind working from a distance. “This work is very gratifying for people like us, who are usually quite isolated in front of our computers here,” she says. “Malaria is a major killer, especially of children, and this is a challenging scientific problem. It's very satisfying to think that we might have an impact on understanding the biology of the disease, and on how it's treated.”