A new chip with a capability to read genomes uses arrays of tiny pH sensors on an integrated circuit to sequence DNA without the need for optics.
It has been six years since massively parallel sequencing was introduced, and a recent report in Nature by Jonathan Rothberg and colleagues of Ion Torrent Systems proves that our appetite for low-cost, large-scale technologies is not on the wane. Rothberg, a pioneer of next-generation sequencing, believed that “we needed a more direct way to convert biological information into digital information.” The historical limitation, in his view, has been the need for sequencing platforms to image at the detection step.
To sidestep the imaging requirement, Rothberg and his colleagues focused on a byproduct of DNA polymerization, a single hydrogen ion. Their key insight was in using an ion-sensitive field-effect transistor, which translates changes in pH from proton release to changes in current, to underlie each microwell in a high-density array. The group fabricated the integrated circuit using the complementary metal-oxide semiconductor (CMOS) process, the same technology used to construct microprocessors and digital camera sensors.
As in other next-generation sequencing methods, the 'ion chip' needs to start with millions of short, isolated DNA sequences, each at high-enough copy numbers for detection. The system uses emulsion PCR, whereby fragmented DNA is ligated to adapters, fixed at one end to 2-micrometer acrylamide beads and clonally amplified in individual droplets of an oil emulsion. Sequencing reactions are kept isolated by the 1.2 million wells etched on the chip's surface, each measuring 3.5 micrometers and only accepting a single bead.
To sequence, each of the four DNA bases is successively flooded across the chip and then washed away. The process takes 4–5 seconds and is repeated over 100 times. Protons are only detected when a base is added, with repeated bases yielding a current proportional to the number of added bases. In this way, the ion sensors take a chemical picture of incorporation events for every nucleotide. Rothberg likens it to “the CMOS imager, which converts photons to electrons and allows us to see light—only we created a chip that 'sees' ions.”
The advantages of ion sequencing are speed, cost and scalability. Its simplicity keeps the footprint small and the costs low because imaging and special reagents such as fluorescently labeled nucleotides are avoided. An average run of 100-base-pair (bp) reads takes only 2 hours and produces 25 Mbp of sequence. Rothberg says that his group is currently sequencing 200 bp routinely and has broken 300 bp. To demonstrate scalability, they produced 6.1-million-sensor and 11-million-sensor chips by increasing surface area and dropping the number of transistors.
Ion Torrent Systems, the company founded by Rothberg and now part of Life Technologies, made this technology commercially available in the form of their Personal Genome Machine late last year. In the current work, they show that it is capable of whole-genome sequencing. They first produced robust sequence data for three bacterial genomes at five- to tenfold coverage. Accuracy lagged somewhat behind that of other methods for the first 50 bp, but was greater for sequences over 100 bp. One limitation was single-base repeats, for which data accuracy drops the longer they are. Using one thousand 1.2-million-sensor chips, the researchers produced a low-coverage personal genome sequence of Gordon Moore, author of the eponymous law, which states that the number of transistors on an integrated circuit will double every two years. Although the results were roughly comparable to those for the same genome sequenced at low coverage on a different platform, ion sequencing is currently best suited to small genomes and diagnostic detection of variants in targeted stretches of DNA.
As a young technology, accuracy and sequence yield are expected to improve. In addition to their own tweaks, Rothberg is banking that innovations in CMOS chip fabrication will also drive improvements. As he puts it, “the publication used a factory built using technology from 1995, but we can use accumulated Moore's law and time travel by making our next chips in a 2005 factory.” He expects it will be possible to make a 1-billion-sensor chip capable of rapidly sequencing personal genomes in the near future.
Rothberg, J.M. et al. An integrated semiconductor device enabling non-optical genome sequencing. Nature 475, 348–352 (2011).
About this article
DNA immunoprecipitation semiconductor sequencing (DIP-SC-seq) as a rapid method to generate genome wide epigenetic signatures
Scientific Reports (2015)
Electronics Letters (2012)