Ion Torrent's Ion 314 semiconductor sequencing chip will initially generate at least 100,000 reads per run with an average length of 100 base pairs. (Click to enlarge)

With a price tag of $50,000, about one-tenth that of its higher-throughput next-gen cousins, the PGM could “democratize” next-gen sequencing, according to the Broad Institute's Nusbaum. “There are a lot of people in the world who don't have access to ‘heavy lifting’ sequencers,” Nusbaum says. “They simply have no opportunity to raise the money to buy an Illumina or SOLiD machine, and even if they did, they wouldn't be able to support it on the budget that they have.”

But it's about more than just price, says Nusbaum; by enabling small research groups to sequence on demand, the PGM would allow almost anyone to become a mini-sequencing core facility. “The democratization isn't so much of a democratization about price, it's democratization of access.”

The system doesn't provide complete democratization, of course: Even the 316 chip won't produce enough sequence for applications such as whole human genome sequencing, says Nusbaum, whose facility is beta-testing the device. Instead, the system is appropriate for specific small-scale applications like microbial genome sequencing, microbial transcriptome analyses, and targeted and amplicon sequencing. At the Broad, the Institute's four PGMs are used for library quality control and accelerating technology development cycles. “Why do I care about this machine right now?” Nusbaum asks.

“The short answer is that it's fast—very fast.” In just a couple of hours, he says, and for just a few hundred dollars, researchers at the Broad can quality-test sequencing libraries that normally take a week and a half or longer—as well as thousands of dollars—to run.

“It's not glamorous, but it's something that the high-throughput centers really want to do,” he says. “We spend a lot of our money sequencing really complex pools, so everything we do, every step that we have, we want to put quality control in place. This is a good thing for that.”

Sequencing's Electron Microscope

Other in-development methods dispense with the sequencing-by-synthesis approach altogether. One alternative—sequencing single molecules directly using transmission electron microscopy (TEM)—is being developed at Halcyon Molecular in Redwood City, CA and ZS Genetics in North Reading, MA.

It won't be easy. The technique's resolution is sufficient to resolve individual DNA bases, but DNA is more or less transparent to TEM; the nuclei present in nucleic acids simply aren't dense enough to reflect the instrument's electron beam. EM-based sequencing therefore requires a contrast agent. ZS Genetics uses an enzyme-driven DNA amplification step to incorporate high–atomic number (Z) labels, distinguishing bases by their differential contrast in the EM. Halcyon uses chemical reactions to incorporate such high-Z agents as Os-BIPY and “monofunctionalized cisplatin variants,” tagging molecules with one label at a time and leveraging base complementarity rules and bioinformatics to stitch the resulting single-base patterns into a complete sequence, says the company's Chief Visionary Officer, Kent Kemmish.

An artist's representation of DNA passing through a grapheme nanopore. (Click to enlarge)

The resulting sequence reads could be among the longest of any method, says Church, who advises 18 next-generation sequencing players. Kemmish says Halcyon is targeting reads in excess of 100,000 bases apiece. ZS is aiming for the 10,000–20,000 nucleotide range, says president William Glover. “We think we will end up with a system that's taking 3–10 [images] per second, and will give 10–50 megabases per hour—and that would be one of our earlier systems,” Glover says.

At those lengths, EM-derived sequences could enable de novo sequencing even of highly repetitive regions that traditionally have been refractory to such methods, says Glover—everything from a 4-kb stretch of the herpes simplex virus to the vast major histocompatibilty region on human chromosome 6. The problem with current short-read technologies, Glover explains, is that eukaryotic repeat elements typically are longer than the reads, making de novo assembly impossible. A long-read technology, though, can collect enough sequence to span the repeats, thereby enabling unambiguous alignment. “If you could give me thousands-of-bases-long reads of even modest accuracy, those would be very powerful for looking at genome structure,” Nusbaum says.