Digital DNA. Novel semiconductor chips enable electronic DNA sequencing.

J. M. Rothberg et al., Nature, 475 (21 July 2011)

A $1000 Genome by 2013?

In 1965, Gordon Moore, co-founder of Intel, famously observed that the number of transistors on computer chips were doubling every 2 years. That rule of thumb since became known as Moore's Law and continues to drive the improvement of computers and other electronics such as the number of pixels in digital cameras and the size of computer memory. Now, Moore himself has become a hallmark in perhaps the next killer application to bend to the electronics revolution.

Researchers at Ion Torrent, a Guilford, Connecticut, company that makes benchtop DNA sequencers and is now a part of Life Technologies in Carlsbad, California, report online today in Nature that they've used their newly developed electronic gene sequencing chips to sequence Moore's genome. The sequencing chips are made with much the same semiconductor technology that has driven the improvements in computing over the past 5 decades. So researchers expect that the new sequencing technology will be able to ride that same wave of technological advances to drastically reduce the cost of sequencing DNA.

The price tag of DNA sequencing has been dropping steadily since researchers completed sequencing the first human genome in 2003 at a cost of about $3 billion. Today, new sequencing technologies have brought the figure close to $1000 per genome, although precise costs are hard to come by. Most sequencing technologies tick off the sequence of DNA's four chemical letters—A's, G's, C's, and T's—by linking fluorescent molecules to the letters and the progression of colors produced by running through the sequence of each DNA strand. This so-called optical sequencing has proved to be a reliable technique. But the optical scanners and imaging software needed to interpret the millions to billions of colored blips and convert them into a gene sequence keep costs high.

Four years ago, Jonathan Rothberg, Ion Torrent's founder, decided to try to replace fluorescent detection of DNA's chemical letters with direct electrical detection. He and his colleagues succeeded quickly, selling their first electronic genome readers last December. In the past 7 months, the technology has become a commercial hit and is already one of the top-selling sequencing technologies on the market.

In their paper, Rothberg and his colleagues detail for the first time a close-up of how the technology works. The current reader consists of a silicon chip patterned with 1.2 million sensors. Each sensor consists of a tiny well, inside which sits a fragment of single-stranded DNA linked to a small bead. Into each well, researchers insert a copy of a DNA polymerase protein that reads the letters in a DNA strand and uses that information to build a complementary strand, in which A's bind to T's, and C's to G's. Each time the polymerase adds a new A, G, C, or T, it spits out a positively charged hydrogen ion; the charge of the ion slightly alters the pH of the solution and can thus be read out by an electronic pH sensor.

To sort out the sequence of letters in each strand, the sequencing machine first floods all the wells in the chip with copies of the A base. In the wells in which the sensor detects a hydrogen ion, the ion shows that the DNA that the DNA polymerase added an A to its growing strand. All the wells are then washed, C's are added, and the cycle continues over and over until all the sequences are identified.

In addition to sequencing Moore's genome, which required 1000 chips that sell for $99 apiece, Rothberg and his colleagues report that they've used their new sequencing chips to decipher three bacterial genomes. And last month, researchers in Germany and China reported that in just 3 days, they had independently used Ion Torrent's chip readers to decode the Escherichia coli strain that recently killed more than 30 people in Europe.

Because the new electronic DNA readers don't require optical image processing to sequence DNA, "it makes it extraordinarily fast," says Jeremy Edwards, a molecular geneticist at the University of New Mexico, Albuquerque. Edwards adds that compared with rival sequencing technologies, the chip approach still has room to improve its accuracy and overall throughput.

But Rothberg notes that the technology is already improving rapidly by taking advantage of the decades of advances in semiconductor manufacturing. The company's first chips, he says, were made in a semiconductor fabrication facility from 1995, in which the smallest features that could be inscribed on chips were 0.35 micrometers across. By simply switching to a 2005 semiconductor fabrication, the company can produce chips with features just 0.065 micrometers, or 65 nanometers, across, enabling them to cram more than 1 billion ion sensors on each chip. Following this path, Rothberg says the company will be able to sequence complete human genomes for $1000 by 2013: "It's so scalable ... the $1000 genome is inevitable."