n 1998 Pål Nyrén and his colleagues at the Royal Institute of Technology in Stockholm, Sweden, published an entirely new method for reading DNA (M. Ronaghi, et al., Science, 281[5375]: 363-5, 1998). That technique is at the heart of the #2 paper, which promises to shift genome sequencing up a gear.

The Swedish technique is called sequencing by synthesis, or SBS, and it uses an entirely different approach to the method that is the basis of almost all sequencing. Fred Sanger's method involves stopping the synthesis of a stretch of DNA at random points in the sequence, sorting the strands by size, and then determining the base at which sequencing stopped.

SBS relies on the fact that when the enzyme DNA polymerase adds a new base to a growing strand of DNA, the reaction releases pyrophosphate. An enzyme called ATP sulfurylase converts pyrophosphate to ATP. And luciferase, an enzyme from fireflies, causes luciferin to emit light in the presence of ATP. The key conceptual breakthrough is the addition of a fourth enzyme, apyrase, that slowly breaks down nucleotides being added to the chain and the ATP produced, effectively cleaning the reaction. Now the way is clear to wash carefully timed waves of DNA letters—A, T, C, and G—past the strand of DNA. If a wave of A creates a flash of light, there is an A at that point of the sequence, and so on. The apyrase gets rid of the excess A and any ATP produced before the next wave—say, T—comes along. Successive waves synthesize the sequence.

The paper at #2, from a huge team led by Jonathan Rothberg of 454 Life Sciences in Branford, Connecticut, automates the process and ushers in sequencing rates 100 times faster than current Sanger technology. Many factors contribute to the end result. The DNA is prepared by breaking it into random pieces and attaching a short adaptor sequence to the ends of each fragment. The adaptors allow the DNA to bind to tiny beads, about the diameter of the finest flaxen human hair. Conditions are arranged so that each bead binds a single piece of DNA and each bead is then encased in a droplet of oil that also contains all the reactants needed for DNA polymerase to do its job of building a new strand of DNA. The oil droplets are part of an emulsion that keeps each droplet and its bead distinct, and the whole lot goes through polymerase cycles to create 10 million copies of the DNA on each bead.

The beads are now put into the wells of a plate. The standard Sanger plate contains 96 wells; Rothberg's group has 1.6 million wells on a slide 6 x 6 cm. The slide is made by packing together glass optical fibers, fusing them, drawing the block out to decrease the diameter of the fibers still further, and repeating. The end result is a glass slide that contains almost 500 fibers per square millimeter. Each fiber is surrounded by a cladding coat. The glass is etched to leave a small pit, not quite twice the diameter of the beads.

The emulsion is broken and the DNA bound to the beads denatured to create single-stranded templates for synthesis. Each bead goes into one of the wells—and only one bead fits in each well—along with other, smaller, beads that carry the enzymes for pyrophosphate sequencing. The slide is clamped to a light detector and waves of bases flowed over the top of the wells. A well that lights up reveals the letter at that point in its DNA's sequence.

The technique reads only about 100 bases, compared to the average Sanger read of about 700 bases, but more than makes up for it by reading 1.6 million sequences at a time. The system, which fits in a cabinet about the size of a microwave oven, read the 580,069 base sequence of a known bacterium with 99.96% accuracy in a single four-hour run. Not counting the computer time to assemble and check the sequence, the entire process took three days as opposed to more than three weeks. Sequencing may be outstripping Moore's law.