NEW YORK – Ultima Genomics, a DNA sequencing startup that surfaced this week, could shake up the top end of the sequencing market by early next year.
On Monday, the day before announcing itself and a cumulative $600 million in funding, Ultima posted a BioRxiv preprint detailing specifications of its novel sequencing-by-synthesis (SBS) method. With 10 billion reads per run, 300 bp single-end reads, a 20-hour run time, and claims of being able to deliver a $100 human genome — the genomics equivalent of the four-minute mile barrier — the company could have Illumina looking over its shoulder.
"This seems to be a real head-to-head competitor to the NovaSeq," said Chris Mason, a sequencing technology expert at Weill Cornell Medicine who has received data from samples run by the company on the Ultima instrument. "The $100 genome at 30X coverage is a pretty disruptive feature. We've generated methylation data, and it looks really good. The data on SNPs look really good, comparable or equivalent to Illumina."
The longer read length, compared to 150 bp pair-end reads for Illumina's high-throughput runs, could be a key differentiator if Illumina is able to match it on price. "If you have the choice, in almost every instance you want to have a longer, contiguous molecule," Mason said.
Lowering the cost of sequencing is the raison d'être for Ultima, according to CEO and Cofounder Gilad Almogy. "When I looked at sequencing, it became obvious, at least to me, that it's a data problem," he said. "If you can generate data at lower cost, there will always be a need for it."
Almogy, who has startup and management experience in the solar and semiconductor industries, teamed up with former Illumina engineer and Genapsys VP of product development Mark Pratt to understand where the costs were and how to address them. They founded Ultima in 2016 and spent the next few years with an "inward focus" on developing their solution, which Almogy describes as "mostly natural sequencing-by-synthesis."
Ultima says it cuts costs where it counts the most: in the flow cell and reagent usage, specifically the polymerase.
Traditionally, SBS flow cells have employed a glass plate sandwich design, where reagents are run through channels and into the proximity of DNA clusters. But a larger flow cell requires a thicker channel and more reagents, Almogy said. Ultima has designed an "open" flow cell based on a silicon semiconductor-style wafer, "generating a dense array of electrostatic landing pads to bind clonally amplified sequencing beads, which are separately produced by an automated emulsion PCR process," according to the preprint. Not only is the wafer cheap, it only requires a thin layer of reagents, allowing for high utilization, Almogy said.
Moreover, Ultima's chemistry is more efficient, he said, decreasing both cost and run time. The firm doesn't use terminators on its nucleotides, like Illumina, which means it not only skips the unblocking step, but also doesn't need to replenish the polymerase or other reagents. The method uses only one flavor of fluorescent label, running each base separately.
Additional run speed is gained by using the round wafer instead of a rectangular flow cell. "Field of view is never wide enough," Almogy said. With a rectangular design, the camera must scan back and forth, accelerating and decelerating at the turns. "Because we're using a circle, we scan it the same way a hard drive does, by rotation. The result is you can enjoy the ultra-high speed of the camera."
The instrument, named the UG 100, will run two flow cells at a time. "One is doing chemistry, while one is doing imaging," Almogy said. "In 20 hours, we're finishing two of those flow cells." Ultima declined to provide a list price for its instrument at this time. "It's a high-throughput box, so it'll be roughly what you'd expect a high-throughput box to cost," he said.
At launch, Illumina's NovaSeq 5000 and 6000 sequencers carried list prices of $850,000 and $950,000, respectively. Currently, a run of an S4 flow cell in 2x150 bp paired-end mode, the highest output available on NovaSeq, equates to about $6 to $7 per Gb.
Ultima is targeting its commercial launch for the first quarter of 2023 and has already started an early-access program with 10 partners at genome centers, pharma companies, and clinical labs. They've managed to attract the attention of the Broad Institute, one of the most important genomics research centers in the US.
The firm worked with the Broad Genomics platform to sequence over 200 whole human genomes and produce the preprint paper. It also partnered with the Broad's genome analysis tool kit (GATK) developers to provide bioinformatics support for the 300 bp, single-end data type. "GATK is adapted for this," Almogy said. The Broad Institute said it did not have anyone available to comment for the story.
Ultima has also partnered with others to demonstrate the viability of its platform in many of the most important high-throughput sequencing applications. Specifically, it teamed up with Aviv Regev of Genentech and Joshua Levin of the Broad for single-cell transcriptomics; with Michael Snyder of Stanford University for whole-genome methylation sequencing; with Jonathan Weissman of the Whitehead Institute for high-throughput perturbation screening; with researchers at Baylor College of Medicine for Hi-C chromosome conformation analysis; and with scientists at Cancer Research UK and the New York Genome Center for tumor profiling.
"We went one by one to make sure there's a solution," Almogy said, "from library prep to bioinformatics." The firm promised more information on these applications in forthcoming preprints and at next week's Advances in Genome Biology and Technology meeting.
With respect to bioinformatics, "not everything is plug and play, but they're all solvable," said Doron Lipson, Ultima's CSO. "We've done a lot of work to make sure these applications are all doable with our data." Lipson noted that the firm has a method to re-index existing sequencing libraries for Illumina and other systems to run on Ultima's platform.
If there's one area where Ultima's platform currently does not meet the standard set by Illumina, it's for homopolymers, or sequences of consecutive bases of the same type, Mason said. "If you get 10- or 11[-base] homopolymers, it does have some errors, which probably lead to lower indel accuracy, compared to NovaSeq." Indel accuracy was approximately 96 percent, compared to about 97 or 98 percent for Illumina. "For a first foray into the market, it's actually very impressive," Mason added.
Lipson said that the system’s performance in regions with GC content above 70 to 80 percent shows "a bit of a drop."
But for many aspects, it is at least on par with Illumina, according to the firm. At least 85 percent of bases are above Q30, or 99.9 percent accuracy, and sequencing the Genome-In-A-Bottle reference sample HG001-7 showed SNP accuracy of 99.6 percent.
"We think we're going to be even better than that," Lipson said, noting that high SNP accuracy is especially valuable in calling rare variants in cell-free DNA data and in minimal residual disease monitoring.
The company also believes it has a path to improve its indel calling. "With new chemistry and better data, with our R&D runs, we're getting to 99 percent [accuracy]," Lipson said. The firm has partnered with Google DeepVariant and says it is the first sequencing company to take advantage of additional input channels to improve accuracy.
"When you look at sequencing data, there are the reads, but there's additional information about the reads," Lipson said. "We output more information about each base," usually quality information, including confidence scores. "Because we have richer quality info, we can use that to make better calls."
Almogy wouldn't give a prediction but suggested that Ultima could push read lengths a bit more. "We know for certain we can go longer," he said, noting that 300 bp is simply the mode for the distribution of read lengths. In the preprint, the authors presented data showing modal read length of greater than 400 bp. However, the goal is not to become a long-read sequencing platform, he said.
The company is focused on launching a high-throughput sequencer but said it may also consider an instrument for the mid-throughput layer of the market. It has 370 full-time employees and "is growing rapidly," a spokesperson said.
Already, orders are coming in. Mason said his lab wants one. And with its lower running costs, Ultima is in discussions to sell globally. "Is this a solution for the whole world, and not just high-revenue, high-GDP countries? I think the answer is, 'Hell yeah,'" Almogy said. "I'm not claiming that what we've done is enough, but it's a step in the right direction."