General OverviewLife Technologies expanded their NGS portfolio with the acquisition of Ion Torrent and their first system, the Ion Personal Genome Machine (PGM) sequencer, in 2010. While this technology uses a sequencing-by-synthesis method and emulsion PCR (emPCR) similar to other platforms, it differs in that it doesn’t use fluorescence or chemiluminescence. Instead, it measures the H+ ions released during base incorporation. The lack of any optics has allowed Ion Torrent to rapidly expand the output from their systems by approximately 10-fold every six months. This rapid pace of improvement, along with fast runs of about 2 hours and inexpensive consumables has made Ion Torrent a popular new platform.
The initial chips generate far fewer reads and much less output than the likes of HiSeq or SOLiD, so the focus so far has been on small genomes and targeted sequencing. However, they have announced a new system and new chips that will allow them to push into the high throughput territory of whole genome sequencing by the end of the year.
Ion Torrent’s first instrument, the Ion Personal Genome Machine (PGM), is the least expensive next generation sequencer on the market, with a list price of approximately $50k and individual runs costing in the range of $300 to $750. Given its output capability (currently up to 1Gb) and speed of runs (2 hours), it is being targeted towards smaller genomes and targeted sequencing. It uses disposable chips which come in three varieties of increasing output (see table below). While the read length (currently at 200b) will almost certainly improve over time, it appears that the number of sensors (and, therefore, reads) has likely maxed out.
In September 2012 Ion Torrent launched their follow on system, the Ion Proton, which allows for larger chips with higher densities needed for exome and whole genome sequencing. The Ion Proton is substantially more expensive at $149k, but is capable of generating much larger outputs. The first chip, the PI, is able to generate ~10Gb per run while the PII (currently slated for early 2015, after several schedule slips) should eventually be able to generate ~100Gb per run (but it should be noted that they have downgraded their claims to ~30Gb per run at launch). Ion Torrent has even announced the PIII chip which will double the number of sensors, generating ~200Gb per run. However, given the difficulties they've had releasing the PII (and the lack of any progress updates), it is unclear when the PIII might be launched.
|PGM 314||PGM 316||PGM 318||PI||PII
(est. early 2015)
|# of sensors||1.2M||6.1M||11M||165M||660M|
|Total output||up to 100Mb||up to 1Gb||up to 2Gb||~10Gb||~32Gb (at launch)|
|Run time||2-4 hrs||3-5 hrs||4-7 hrs||2-4 hrs||2-4 hrs|
|Output/day*||up to 200Mb||up to 2Gb||up to 4Gb||~20Gb||~64Gb|
|Avg read length||up to 400b||up to 400b||up to 400b||up to 200b||100b|
|# of single reads||up to 0.6M||up to 3M||up to 5.5M||up to 82M||up to 330M|
** template prep plus sequencing per chip
|PGM 314||PGM 316||PGM 318||PI||PII
(est. early 2015)
|Human Whole Genome|
Technology DescriptionThe workflow consists of four major steps: library construction, template prep, sequencing and analysis.
1. Library Construction
The first step in the workflow is library construction. While there are a number of options available, the process is fairly standard and generally involves taking DNA (or RNA converted to DNA), fragmenting it to a uniform size (generally 200-400b) and then adding sequencing adapters.
Ion Torrent has specific kits for a variety of applications, including DNA fragmentation for small genomes, total RNA-Seq and cancer-specific targeted sequencing.
2. Template Prep/Amplification
The fragments generated during the library prep are attached to beads and amplified using emulsion PCR (emPCR). Beads coated with complementary primers are mixed with a dilute aqueous solution containing the fragments to be sequenced along with the necessary PCR reagents. This solution is then mixed with oil to form an emulsion of microdroplets. The concentration of beads and fragments is kept low enough such that each microdroplet contains only one of each (or possibly none, but almost never more than one). Clonal amplification of each fragment is then performed within the microdroplets. Following amplification the emulsion is ‘broken’ (generally by organic extraction and centrifugation) and the amplified beads are enriched in a glycerol gradient (with unamplified beads pelleting at the bottom).
While the emulsion PCR process is effective, it is slow and complicated. Ion Torrent is working to improve emPCR, first through the release of the Ion Xpress™ kit, which reduced the overall time from 6 hours to 3.5 hours. They have also launched their “Ion OneTouch” system, which automates much of the process, reducing hands on time to a few minutes per sample.
Ion Torrent has also announced improved template prep machines for the first half of 2013, the OneTouch 2 and the Ion Chef. Further out, they have promised the Avalanche system, an emulsion PCR-free isothermal amplification system, slated to be ready in early 2014 with the PIII chip.
What really differentiates Ion Torrent’s systems is the sequencing technology. It is based on the standard pyrosequencing chemistry, a form of ‘sequencing by synthesis’ whereby individual bases are introduced one at a time and incorporated by DNA polymerase. However, unlike other platforms based on pyrosequencing, rather than measuring light released from chemiluminescent reagents, the Ion Torrent system measures the direct release of H+ (protons) from the reaction. Because optics aren’t required, they’re able to make relatively inexpensive instruments coupled with disposable chips, which essentially act as pH meters. The lack of optics also means they don’t have to contend with slow image scans, so the sequencing reactions are relatively fast, with 200b reads taking about 2 hours. Finally, the lack of fluorescence or chemiluminescence means that the system can use unmodified nucleotides, which are cheaper and better tolerated by DNA polymerase. While the error rates are generally pretty good (~1%), pyrosequencing chemistry has trouble with long homopolymers (stretches of the same base, e.g., AAAAAA). Because the chemistry doesn’t pause after each base incorporation, stretches of the same base will result in a single, albeit stronger, signal. While short stretches can be differentiated, it becomes increasingly difficult with longer stretches.
4. Data analysis
As the Ion Torrent systems generate standard output files like FASTQ, data analysis is generally straightforward. In addition to a variety of available third party analysis solutions, Ion Torrent offers the ‘Torrent Browser’ software, which acts as the primary interface for a number of basic functions. Later this year, they will be launching a cloud-based solution called ‘Ion Reporter’ that will serve as a front-end for a variety of open source analysis solutions.