wo beasts, two genomes; between them they cause hundreds of millions of cases of malaria and kill more than a million African children each year. The sequences of Plasmodium falciparum, the parasite that causes one of the worst types of malaria, and Anopheles gambiae, the mosquito that transmits it, sit at #9 and #12 (the latter is R.A. Holt, et al., Science, 298[5591]: 129, 4 October 2002; 33 citations this period), for all the world like just two more examples of DNA data brought to light. But they are more than that, and not just because they are so destructive of human lives.

Almost all of the commentary that surrounded the publication of the two genomes, in October 2002, addressed questions of public health and priorities of scientific and medical research. Tough questions, too. Such as, would the money spent decoding the DNA of the two species have been better spent on simple prevention and cure? Can mosquitoes be genetically modified to break the chain of transmission? Already achieved in a laboratory model of malaria transmission in mice, would such a strategy work for P. falciparum worldwide, and what might the long-term consequences of this approach be? Will the genome genuinely offer new targets for therapeutic drugs? How soon? The debates on these and other matters continue to rage, with no satisfactory conclusions, and threaten to obscure some of the scientific breakthroughs that the two sequences represent.

The parasite, for example, is the most AT-rich genome sequenced to date. More than 80% of the bases are As or Ts, with less than 20% Cs and Gs. That posed a huge obstacle to completing the sequence. Like so many other sequences, that of P. falciparum was attacked by breaking the DNA into millions of fragments, sequencing each of those, and then relying on computer power to assemble the fragments into a coherent whole. In this procedure the software looks for identical sequences at the ends of different fragments, on the assumption that those fragments will then be neighbors in the genome. But if As and Ts are so much more prevalent than Cs and Gs—with runs of 50 or more As and Ts common—how is the software to detect matching regions to assemble the fragments into longer sequences?

The answer was to derive new kinds of maps. One was an "optical" map. The researchers cut each Plasmodium chromosome with a restriction enzyme that breaks the DNA at a known target sequence. Then they sorted the fragments by size. Then they took the putative sequence and asked the computer to break it at the same target sequence. This generated a set of virtual fragments that could also be sorted by size, which could be matched against the real optical map. The virtual fragments could then be juggled and re-assembled until there was a close fit between the real and virtual maps.

The need to solve the problem posed by the AT imbalance was one reason why the Plasmodium genome, reasonably small at only 23 million base pairs, took some six years to complete. By comparison the Anopheles genome, more than 10 times bigger, was reasonably simple to assemble. The big problem there was that the chosen strain of mosquito, the so-called PEST strain maintained by the Institut Pasteur in Paris, proved to be more diverse than expected. This may be because two distinct populations of mosquito originally contributed to the PEST strain, which was created by crossing wild mosquitoes with a laboratory strain that carried a mutation for eye color, enabling researchers to track changes in parasite and host.

Different pieces of sequence were derived from different individuals. Because those individuals were more distinct than previously thought, assembly of adjacent stretches of DNA proved difficult. Were mismatches between overlaps real, or were they instead a reflection of the differences between the individuals from which they had been derived? In the end the Anopheles sequencing team turned to a detailed analysis of previously made artificial chromosomes, one reason why the PEST strain had been chosen in the first place, to ensure that the final assembly reflected the true sequence.

Both genomes also contain the usual insights into their owners’ way of life that one has come to expect from sequencing efforts. The mosquito, for example, has a large number of genes dedicated to sniffing out the presence of a suitable (human) host. The parasite has several genes for proteins that are expressed on the outside of infected red blood cells, causing them to stick in smaller blood vessels. The host’s immune system recognizes and attacks these proteins, but the genes are near the end of the parasite’s chromosomes where they can recombine and shuffle easily, helping the parasite to evade the host’s defenses. Both may ultimately prove suitable targets for interventions designed to blunt the ability of Plasmodium falciparum to scythe down millions of people each year.