10X Genomics at AGBT
By Aaron Krol
February 25, 2015 | The Advances in Genome Biology & Technology conference kicked off today, as hundreds of scientists and investors descended on the Marco Island Resort off Florida’s Gulf Coast for the premiere annual event in the world of commercial genomics. The AGBT conference has a reputation for splashy announcements — memorably including Oxford Nanopore’s somewhat premature unveiling of the handheld MinION sequencer, a device that has come at least two years too late but not, perhaps, too little. This year, however, the major sequencing companies seem unlikely to make any big power plays at the event: Illumina already laid out its new product line at the JP Morgan Healthcare Conference in January, Thermo Fisher’s P2 Chip is trapped in development, Roche and QIAGEN have set long timetables for their promised sequencing portfolios, and while PacBio’s gold sponsorship of AGBT has added fuel to rumors of a new benchtop instrument, the company seems most eager to use the event as a showcase for the many unique applications of its long-read data, a strategy that has won the company some crucial niches in a market where others have struggled to make inroads against omnipresent Illumina.
That doesn’t mean that there won’t be exciting new products parallel to sequencing, though. One new company hoping to make an impact this year is 10X Genomics, who shared some tantalizing details with Bio-IT World this January about a technology suite built to work in tandem with short-read sequencers to recover long-range genomic information. 10X has scooped up a bronze sponsorship at AGBT, and will be unveiling a full commercial version of its technology, together with lots of performance data collected both internally and with early access customer Hanlee Ji of Stanford School of Medicine, who will speak on Saturday about using 10X products to uncover structural variants and haplotypes in complex cancer samples.
“It’s tremendously exciting for me personally,” Serge Saxonov, the CEO of 10X, told Bio-IT World. “We’re really looking forward to this week and showing [our platform] to the rest of the world.”
With today’s official launch at AGBT, the 10X technology finally has a name: the GemCode Platform, including an instrument, chemistry kit, and informatics software. The platform is named for the “gems” into which DNA samples are partitioned in the instrument through a custom microfluidic system. Users insert their DNA samples onto a GemCode Chip, along with gel beads and a partitioning oil, and inside the instrument’s microfluidics, the oil naturally separates long DNA molecules — which can extend to 100 kilobases or longer — into different gems. The DNA in each gem can then be fragmented into short-read libraries suitable for Illumina sequencing, with each fragment receiving a 14-base molecular barcode unique to its gem of origin.
The GemCode software will later use those barcodes to do something an Illumina sequencer can’t do on its own: link together short reads that occur kilobases apart from one another. “Since our technology barcodes every fragment that makes it on the sequencer, we always know the original molecule from which those reads emanated,” says 10X Vice President of Marketing Rob Tarbox.
The GemCode Instrument and associated products. Image credit: 10X Genomics
The instrument is a fairly hands-off addition to a normal library preparation workflow, with just a few minutes of sample loading per run. The partitioning model, in which an oil is used to physically separate out single DNA molecules from a sample as they pass through microscopic channels, also allows GemCode to work with minute quantities of DNA: according to 10X, as little as one nanogram of sample material. (That system has also quickly attracted a lawsuit from RainDance Technologies, which uses the same principle in its single-molecule PCR devices. Responding to RainDance’s patent infringement suit, Saxonov told Bio-IT World, “We’re firmly of the view that nothing we’re doing is infringing any of those claims, and to the extent that they might attempt to extend those claims, we believe that those claims are invalid.”)
The upshot of the GemCode DNA treatment is that scientists using both the GemCode Platform and an Illumina sequencer should be able to get much the same information about large structural elements that they would with a long-read instrument, at a fraction of the upfront investment. 10X earlier advertised that their instrument would sell for $75,000, and they’re now estimating that the price of consumables will run somewhere in the range of $500 per sample.
To date, no technology can reliably capture multi-kilobase structural variants, phase haplotypes, or perform de novo assembly of large genomes at this kind of price — all applications that 10X or its early access customers have been experimenting with.
“We’re getting really long-range information routinely, very easily, and we’re seeing things I’m not even sure if anyone else is able to see,” says Saxonov. “Which is really cool.”
Linked Read Data
One point of validation that 10X will be stressing at AGBT is its recharacterization of very well-known genomes, such as the human cell line NA12878 and multiple cancer cell lines. “We’ve tried to mine existing data as much as we can to understand the performance of our system,” says Tarbox. “We can see a lot of things that have already been seen in previous studies, but we’ve also gotten glimpses of the interesting things our platform can provide that haven’t been seen previously by the other techniques and methodologies out there.”
To use gene fusions as an example, whole genome and exome sequencing with GemCode barcodes not only replicated gene fusions seen in past experiments, but appear to have uncovered a new gene fusion in the well-studied cancer cell line NCI-H2228 between the ALK oncogene and EML4. In addition, says Saxonov, “Not all of these gene fusions look the same. With the short reads you sort of assume they’re all the same, but there’s tons of context around them. Seeing things that people assumed they were seeing, but actually turn out to be not correct, is tremendously exciting.”
Capturing these structural events requires not just a way of barcoding DNA samples, but software that can interpret the resulting sequence data. With GemCode Software, 10X Genomics has chosen to release a platform that offers both predesigned workflows for typical users, and an open API and development environment for bioinformaticians who want to add their own algorithms to the mix.
For the former group, GemCode includes standard pipelines for calling large structural variants and resolving haplotypes, a genome browser for visualizing the results, and a “matrix view” that highlights the barcode signals used to flag variants. Users can also plug in their preferred variant callers, aligners, and other common pieces of software used to analyze sequence data.
“Out of the gate, we’ve put together these solutions for at least the early applications we see people going after,” says Saxonov. “And we expect people will develop applications on top of that. We won’t be able to address the entire world of applications.”
The closest analogue to GemCode’s computational methods is probably Moleculo, a system acquired by Illumina two years ago that also uses molecular barcodes to recover long-range information on DNA molecules. But the 10X team stresses that Moleculo’s “synthetic long read” technique, of completely reassembling longer fragments from barcoded short reads, is relatively inefficient.
“We don’t reassemble synthetic long reads most of the time, unless our users are really intent on that,” says Saxonov. “You don’t want to sequence the same molecule over and over again to reassemble it. You’d rather sample more molecules, and get a lengthier representation of each of those molecules.” The GemCode Software works instead on the principle of “linked reads,” where every short read can be binned back to its DNA molecule of origin, but that entire molecule is only rebuilt to the extent necessary to resolve structural questions.
10X Genomics, working with its custom microfluidic system instead of standard lab equipment, can afford to be rather more ambitious than Moleculo, setting up a workflow that can process much longer DNA molecules and track over 100,000 different barcodes at once, while still scaling back the amount of hands-on work required. “When you work in plates, it’s an incredibly painful process to get any data generated,” says Saxonov. “In our case, it’s incredibly easy to generate this data, so from being an onerous, unforgiving process… [we’ve created] a straightforward, easy workflow with a massive scale.”
Illumina, whose sequencers are currently the only instruments compatible with the GemCode Platform, is more likely to see 10X as a commercial partner than a threat to its small Moleculo unit. All the major sequencing companies are accustomed to outside groups building out accessories to improve their sequencing data. Granted, this normally comes in the form of a few new open source algorithms rather than a full-fledged product line, and it’s unusual to see an upstart like 10X Genomics graft its whole platform onto a much larger company’s products.
For his part, Saxonov says that he’s been in touch with Illumina representatives since his company emerged from stealth mode in January, and that he expects the company to be supportive of the GemCode Platform. “I think we’re very synergistic,” he says, “because we’re opening up a lot of great new applications for their sequencers, and making a lot of the existing applications much more powerful. Ultimately our presence helps them.”
10X is taking orders for its instruments at AGBT, and plans to start shipping in the second quarter of 2015. So far, all the data from its platform comes from either 10X itself, or early access users who have had easy access to the 10X team for trouble shooting, so this first round of buyers will provide some important perspective on how GemCode performs in the field. With luck, it will represent a real breakthrough for efficiently getting long-range genomic information, an issue with which the genetics community has always struggled.
Of course, 10X also wouldn’t be the first company to fly into Marco Island with big promises, only to fizzle on delivery. Either way, given the high visibility that comes with a launch at AGBT, we’ll know soon enough.